;hemically derived ceramic composites
By
BURTRAND INSUNG LEE
DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN
PARTIAL FULFILLMENT OF THE REQUIREMENTS
FOR THE DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
1986
ACKNOWLEDGEMENTS
It is difficult for me to acknowledge everyone who yave helping
hands during the tenure of my research as a graduate student at the
University of Florida.
It is hardly necessary to mention my advisor Professor Larry L.
Hench for his inspiring support and stimulating thoughts. His sense of
humor maintained a relaxed research atmosphere. He helped me to mature
as an independent scientist. I am especially grateful for the freedom
and independence he has given me throughout this research. li was in-
deed a privilege and pleasure to work with him.
Dr. Robin Sinclair at 3H Co. not only provided the facility to
synthesize some of polysilanes but also taught me the techniques in
polymer chemistry. The analytical services of 3M Co. which provided
analytical data of the polysilanes are acknowledged with gratitude. Dr.
Curt Schilling at Union Carbide is also acknowledged for his support by
providing some of their experimental polysilanes.
My thanks are extended to D. Dunnagan and U. Folz for thermal anal-
yses, E. Jenkins for SEM, W. Acree for XRD, Dr. J. Newkirk and C. Turner
for TEM, Dr. A. Gupta for GC, S. Yoon and S. Kong for proton NMR, Lester
at the Engineering Machine Shop of the University of Florida, G. LaTorre
for FT-IR, S. Kang for density and microhardness measurements, and
Professor Batich and Dr. S. Kurinec for help in XPS.
The financial support of the U.S. Air Force Office of Scientific
Research through contract no. F49620-83C 0072 was certainly an essential
part of my life at the University of Florida and is especially acknowl-
edged as is the encouragement of Dr. D. R. Ulrich, contract monitor.
ii
I am also grateful to the members of my supervisory committee.
Professors C. Baticn, M. 0. Sacks, L. Malvern, 0. Clark, and E. D.
Whitney, for their advice and reading of the entire manuscript of this
dissertation.
On the nontechnical side, special personal thanks goes to my
family. Instead of complaining about not spending much time with them,
they rather were truly the Gatorade, "a thirst quencher." My mother-in-
law, in particular, played too great a role to describe.
m
TA3LE OF CO.\T£NTS
ACKNOWLEDGEMENTS i i
LIST OF ABBREVIATIONS, ACRONYMS, INITIALISMS, AND SYMBOLS vi
ABSTRACT ^ ^
CHAPTERS
I. OVERVIEW OF CHEMICALLY DERIVED CERAMICS 1
II. SILICON CARBIDE FROM 0R6AN0SILANE PRECURSORS
Introduction 1q
Experifnental 15
Results 31
Discussion 54
Conclusions 96
III. SILICON CARBIDE/SILICA COMPOSITES FROM CAR30SILANES
AND ALKOXYSILANES
Introduction luij
Experimental Hj2
Resul ts ] ■ 103
Discussion 142
Conclusions 152
IV. SILICON CARBIDE/SILICA COMPOSITES FROM COMMERCIAL
SILICON CARBIDE AND SILICON TETRALKOXI DE
Introduction 155
Experimental 153
Results 167
Discussion 213
Conclusions 230
V. OTHER CHEMICALLY DERIVED CERAMIC COMPOSITES
Introduction 233
Experinental 234
Results 235
Di s cu s s i on 238
Conclusions 257
TV
Table of Contents (continued)
Page
VI. CONCLUSIONS AND RECOMMENDATIONS 259
RE FE RE NCE S 264
B lOGRAPH ICAL SKETCH 272
LIST OF ABBREVIATIONS, ACRONYMS, INITIALISMS, AND SYMBOLS
Ac Acetate ion or group
AIBN Azobisisobutyronitrile
A-PSS Allylic polysi 1 astyrene
B.E. Binding energy in eV
BPO Benzoyl peroxide
-C=C or '^"y.v Vinyl group
-C-C=C or ~\s Allyl group
CDCl^ Deuterochloroform
DgDg Deuterobenzene
CFRI Chemical free radical initiator
DCCA Drying control chemical additive
DCP Dicumyl peroxide
DMDCS Demethyldichlorosilane; Me2SiCl2
DSC Differential scanning calorimetry
DTGA Derivative thermogravimetric analysis
EDS Energy dispersive x-ray spectroscopy
en Ethyl enedi amine
EtOH Ethanol
FID Flame ionization detector
FT-IR Fourier transform infrared
GO Gas chromatography
GPC Gel permeation chromatography
IR Infrared
J-PSS PSS prepared by Shinnisso Kako Co., Japan
vi
Me Methyl group, -CH3
Hn Number average molecular weight
MeCl2 Methylene chloride
MPDCS Methylphenyl dicnl orosil ane, MeSiPhCl2
Mrad Mega rad
M.W. Molecular weight
PC Polycarbosilane
POMS Polydimethyl silane
PDS Polydimethyl silane
Ph Phenyl group, -C^jH^
PrOH Isopropyl alcohol
PS Polysilane
PSS Polysilastyrene
PSS-0 Oligomer fraction of polysilastyrene
P-PSS Petrarch's polysilastyrene
R Reflectance, or diameter to length ratio of a fiber
RT Room temperature
T Temperature, or transmittance
TEOS Tetraethylorthosil icate: tetraethoxysilane
TMOS Tetramethoxysi lane
TMS Tetramethylsilane
SEM Scanning electron microscopy
SS Silastyrene, oligomer fraction of polysilastyrene
STEM Scanning transmission electron microscopy
TEM Transmission electron microscopy
VI 1
TGA Thermogravimetric analysis
TMA Thermonechanical analysis
UV Ultraviolet
v/o Volume percent
ViSO Vinyl ic silane oligomer
ViSP Vinyl ic silane polymer
w/o Weight percent
w/v Weight percent volume
XPS X-ray photoel ectron spectroscopy
XRD X-ray diffraction
Greek Symbols
a Linear coefficient of thermal expansion
V Uncharged high energy electromagnetic quanta
p Density, g/cc
\ Wavelength
V Frequency
V Wavenumber
A Differential value
a Stress, strength
Subscripts
X Crystalline
g Glass
c Carbide
VI 1 1
Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the degree of Doctor of Philosophy
CHEMICALLY DERIVED CERAMIC COMPOSITES
By
BURTRAND INSUNG LEE
May iy8b
Chairman: Dr. Larry L. Hench
Major Department: Materials Science and Enyineering
Silicon carbide was made from various organosilane precursors by
crosslinking and pyrolyzing then in an inert atmosphere. Crosslinking
of these silane precursors was studied by various means. The most suc-
cessful means of crosslinking was found to be via a chemical free radi-
cal initiator, dicumyl peroxide. The mechanism of crosslinking for the
precursors was determined.
Pyrolyses of the silane precursors were carried out and increased
ceramic yields after crosslinking were shown as compared with uncross-
1 inked precursors. The ceramic yields determined by TGA ranged from 10-
70% depending on the precursors and the crosslinking treatments.
Partially densified sol-gel derived silica monoliths were impreg-
nated with the silane precursors while the silica monoliths were still
ix
highly porous. Diamond microhardness values increased 2-3 times from
unimpregnated gel derived silica monolith. A noderace increase in frac-
ture toughness, Kjq, and flexural strengths was achieved. Optical and
mechanical properties, and porosity data are presented.
Using sol-gel silica precursors and tne processing techniques of
polysilanes to obtain 3-SiC, molecular composites of SiC witn Si02, with
Ti02 and with AI2O3 were made in monoliths and powder forms. Monolithic
composites with a molecularly dispersed SiC phase in the Si02 gel matrix
showed a hardening effect by the SiC phase.
The molecular composite powders of SiC/Al203 showed no or little
crystallization of either phase after heating to 1400°C.
Monolithic silicon carbide/silica composites were made using com-
mercially available fibrous silicon carbides and a tetral koxysi 1 ane
precursor. Modest to low flexural strengths were obtained after heat
treating to 900-1400°C, because of the high porosity in the composite.
Cold pressing of the SiC and silica sol slurry improved the density and
flexural strengths. Notched 3-point fracture toughness values, Kt,-, was
as high as 7 MPa^m-'-'^. Excellent thermal shock resistance and oxidation
resistance of these composites are shown.
CHAPTER I
OVERVIEW OF CHEMICALLY DERIVED CERAMICS
Ceramic materials are of critical importance in high technology
(high-tech) areas where unique combination of properties, such as high
strength, strength retention at high temperature, low thermal and elec-
trical conductivity, high hardness and wear resistance, and high chemi-
cal stability are required. However, because of their brittle nature,
ceramic materials produce problems in design reliability in high perfor-
mance structural applications resulting in catastrophic failures under
stress. This inherent problem combined with poor cost effectiveness in
fabrication of complex shapes severely limits wider applicability of
current ceramic materials.
Ceramic materials derived from chemical reagents have the potential
to overcome these problems by
1) Low temperature processing compared with traditional ceramic
processing,
2) Starting chemical compounds that can easily be purified to in-
crease the purity of the ceramic materials,
3) Having versatility in forming complex shapes and precise control
of each step in the processing,
4) Rendering homogeneous mixing; uniformity and, thus, reliability
of the formed bodies can be improved, and
5) A unique combination of microstructure and phase assemblages not
obtainable by traditional ceramic processes may be obtained.
Ceramic materials obtained by chemical processing are a rather
recent development, despite the fact that the science behind the pro-
cessing existed long before the ceramic applications were realized.
Traditional ceramic science has been based more on physics and has been
developed by optimizing the physical behavior with the Microstruct'jre of
the material s.
The term "ultrastructure processing" of ceramics has been intro-
duced^ to represent the chemical manipulation and control of surfaces
and interfaces during the earliest stages of formation in atomic or
molecular scales. The so called "high-tech" ceramics are largely based
on "ultrastructure processing" as this is one way in which engineering
ceramics can potentially yield properties approaching the theoretical
1 imit.
Organic chemistry, once an anathema to ceramists, is recognized as
a major source of new ceramic materials. By using ordinary chemicals as
precursors to ceramic materials, one can study and control the chemical
process in every step during the evolution of ceramics, from the start-
ing chemical to the final product. Greater versatility in fabrication
with more precise control of the process leading to dn extremely homo-
geneous composite with superior properties is the goal and advantage of
this approach.
Ceramic fibers, optical glasses, ultrapure and ultrafine powders,
and ceramic monolithic parts are some of the demonstrated materials
derived chemical ly.i »2 Figure I-l summarizes some of the chemical pro-
cesses.
Add to powder
as binder
Carbo si lone
Metal organ<c
Metal olkoxide
Metal salts
Composite
ceramtc
body
PyrolySiS or
theffTKil
decomposition
Hydrolysis
Polycondensalion
Impregnate
Porous ceramic
cootinq/ttiin film
Fiber
Powder
Impregnate
Porous ceramic
body
— ^/
Partiotly- polymerized '
sol \
Use OS
binder
Metal carbides
Metal nitrides
Thermal
Treatment
Composite
ceramic
body
Fibers
_L
Muiticomponent
homogeneous
noncrystalline gel
Thermal
Treatment
Monolittiic porous
structural
Of reoctive
gel powder
sintered
body
Sintering
Glass or
Gloss ceramics
Fig, I-l . Flow Diagram of Some of Chemically Derived Ceramics
and Composites
The sol-gel method, as shown in Fiy. I-l, is a notable example of
obtaining oxide ceramics from metal -organic precursors. Pure monolithic
parts, thin coatings, matrices for reinforced composites, etc. have been
produced with controlled properties. Uranium oxide fuels were fabri-
cated by the sol -gel method at Oak Ridge National Laboratory in the
1970's.3 Active research is underway to understand the fundamental
chemistry in the reaction steps, as well as in the appl ications. 2'"+ All
facets of chemistry are involved. For example, a nuclear magnetic
resonance technique has been found helpful in understanding the reaction
mechanism of the sol-gel process. 2
It has been shown2 that certain chemical additives change the phys-
ical-chemical state during sol -gel transformation. The mechanism of how
these additives function chemically is not fully understood. Various
dopants may be added to the sol as a chemical reagent by forming a mo-
lecularly homogeneous solution.
Tne most understood sol-gel process is in the production of silica
glass. Silica sol-gel reactions involve hydrolysis and polycondensation
steps of a metal-organic precursor, as shown in eqs. I-l and l-Z.
($4ijOR)4 + n HgO -.- = Si-OH + 4 ROH
2 = Si-OH > = Si-O-Si = + H2O (1-2)
The hydrolysis and polycondensation reactions initiate at numerous
sites within the Si(0R)4 precursor + H2O solution as mixing occurs.
They eventually form a three dimensional linkage of Si-O-Si in a sub-
micron scale and are called sol particles. The sol particles come in
contact to form a gel network. As the gel network is aged at an ele-
vated temperature, the monolithic body is strengthened and becomes more
like a ceramic body.
Some other oxide ceramic materials derived chemically may be given
below, 1) alumina from A1(0R)3 by Yoldas,5'5 2) lead titanate tiy Gurko-
vich and Blum,'' 3) inaium tin oxide films by Arfsten et ai.,8 4) mono-
sized SiOo and TiOo powders by Barringer et al.,^ and 5) single and mix-
ed phase oxide powders by Mazdiyasni . 1°
Similar to the sol-gel process of obtaining metal oxides, pyrolysis
of organometal 1 ic precursors results in nonoxide ceramic materials of
the constituent elements. Thus far, successful examples are silicon
carbide (SiC) and silicon nitride (Si3N4) from polymers containing sili-
con-carbon and silicon-nitrogen bonds in the backbone. "* Boron nitride
and boron carbide can also be made from organometal 1 ic precursors.'*
Titanium carbide, titanium nitride, titanium boride, silicon boride, and
aluminum nitride may be possible from organometal 1 ic precursors.'* Table
I-l lists ceramic materials that can be made from chemical processing of
organometal 1 ic compounds.'*
It is difficult to make complex shapes of dense refractory ceramics
such as SiC or Si^^f^ using convem:ional high temperature sintering, hot
pressing, or hot isostatic pressing methods without a sintering aid.
Grain boundary phases are often introduced in materials, degrading high
temperature performance and oxidation resistance. Refractory carbide
and nitride fibers are nearly impossible to make using traditional pro-
cessing methods.
In making fibers from pyrolysis of an organometal 1 ic precursor,
densification accompanies pyrolysis, which eliminates a separate sinter-
ing process. By analogy to a carbon fiber made from a carbon polymer,
i_
o
(/)
:_
1)
o
o
^-
o
ro
^^
O
at
r—
+J
•*
<D
>^
£_
+->
o
'r—
ai
(/>
-C
C
1 —
0)
Q
'.C
o
•
•
(NJ
^
IT)
Ln
^
^
o^
m
CM
CM
CM
oo
cv
^
'S-
CO
LD
oo
<=3-
CM
•r-
+->
5
c
cn
OJ 01
1— -a
h-
O)
CU
■CI +->
c: c
=) O)
+J O
O 1—
3 (1)
-o >
O 0)
:_ Q
a.
CV
^ — ^
^ — ^
^ — ^
, — ^
-O
t/1
1 — 1
V
- — .
CM
C_3
^
z:
:a
CO
'-O
«:
C^
r—
■ r-
n
1^
r—
CO
CD
• r-
1 —
■<
1—
a.
<c
(D
■r—
1—
■ ■
■ •
^ ^
::a
— ■
o
tyi
■ ^
• —
a;
^ —
* —
Ol
<v
0)
T3
(U
o;
-^
t:
^
CU
■o
OJ
OJ
T3
•f—
-a
O
•f—
T3
T3
•r—
-Q
t_
L
• r—
•^
t-
•f—
C_
t_
+j
-(-)
-C
P^
Q
t-
c_
o
n3
., —
^
.^
ca
O
O
ca
O
5
ST.
o
VI
c
E
E
E
E
-C
E
3
E
c
3
3
3
3
Q.
3
C
3
o
■ r—
*^-
C
.f—
■ r—
•^-
• r—
u
C
cr
• t—
C
C
C
S
O
•r—
<13
TJ
=
(D
o
03
=s
f^
r^
+->
+-)
3
J->
t_
4->
r—
"3
■r—
■ ^
• r—
r^
• r—
o
• ^
eC
O
(jO
1—
1—
■=c
h-
ca
1—
c
o
■!->
+->
u
3
■o
o
o
t—
o
fO
^^
o
CT>
•r-
.!->
M
OJ
>.
^
4-1
o
<u
lO
-C
c
1—
OJ
o
cv
CM
cv
c^j
CM
E
OJ
o
«:r
,-^
s:
o
m
•1—
•r—
.—- -V
0)
(/)
OO
- — ^
C_5
r^
■* —
■z.
>*
X}
cc
ca
<o
OJ
0)
> — ■
"^ '
^M
T3
■o
"^
•r-
»^
OJ
(U
fO
^
c_
■c
■o
>
t_
•!->
•r-
<
(C
.^
e
ja
o
Z
+j
t_
(/I
.,—
rn
■M
c
c
^
(_I
O
o
o
3
o
o
e
c
■o
■ ^-
• t—
o
o
o
r—
r—
(_
I_
(-
• r—
• r-
o
o
a.
00
oo
ca
ca
<u
■a
to
oa
-H-
organometal 1 ic precursors offer a means to make refractory carbides and
nitrides at potentially lower temperatures with the easy forming opera-
tions of traditional polymers.
One of the primary thrusts in applications of these materials is in
gas turbine engines because these materials are strong and stable at
temperatures no metal can withstand, and they also have high thermal
shock resistance, which is necessary for heat engine components. Cer-
amic materials derived from polymer precursors in the form of foam raay
also be used for thermal insulation, filtration, and packing. Thin
films may be applied in electronic devices and metal -ceramic joints.
Extremely homogeneously doped high temperature semiconductors may be
made this way as well. Boron nitride fiber made from organometal 1 ic
precursors can be used as a dielectric material where alumina and silica
fibers are less desirable. These organometal 1 ic precursor materials can
also be used as binders in powder forming processes. There is also a
high probability of obtaining infrared transmitting films of sulfides
and selenides, superconductive fibers of NbN, NbC, silicides, sulfides,
and borides via polymer precursor pyrolysis.
Intrinsic flaw sensitivity and brittleness continue to impede the
broader applications of monolithic ceramic components. These instrinsic
weaknesses can be overcome by incorporating a high modulus, small diam-
eter ceramic reinforcing phase in a ceramic matrix to change the failure
mechanisms to tough, noncatastrophic modes.
Composite materials on the ultrastructural level can be achieved by
mixing polymer precursors containing constituent elements, e.g., a
polysilane mixed with polyphenyl borazol e yields a composite of SiC/3N.3
Polymers containing Si, C, and N can be used to obtain a SiC/Si3N4 com-
posite.'*
A reaction of polycarbosi 1 ane and Ti(0R)4 can yield a composite of
SiC/TiC.i2 In this composite process Ti(0R)4 not only provides the TiC
phase, but also crosslinks the polycarbosi 1 ane, hence maintaining the
shape of the green body during the subsequent heat treatments and in-
creasing the ceramic yield. The structural scale of these precursor
based composites is in the 1-10 nm range, as compared with the 1 to 100
]im or larger range of composites made by traditional processes. ^3
Composites are leading a new era in structural engineering. The
development of high performance materials and advances in fabrication
technology are laying the groundwork for revolutionary changes in
structural design. In order to go forward with high speed in ceramic
composite technology, it is necessary for engineers to break away from
engineering thought processes that have been developed over decades of
working with conventional materials.
Active research on ceramic matrix composites began no earlier than
1982, according to Persh.^"* Even then, the ceramic matrix composites
were based more on applied physics using the traditional processing
methods, such as hot pressing matrix phase powder with a reinforcing
phase.
The objectives of this dissertation are thus based on an explora-
tory study and development of new methods to obtain ceramic materials
derived by chemical means. Processing and properties of SiC/SiOn
composites utilizing techniques of sol-gel derived Si02 and SiC via
organosil anes are the main topics of this work. Other ceramic compos-
ites from chemical origins are also part of this dissertation. The pri-
mary motivation and objectives for the work presented in this disserta-
tion are an interest in the development of new processing methods based
on chemical processes and an understanding of these processes. For cer-
amic and composite materials in this work, the emphasis is more on con-
cepts rather than the final products with exciting quantitative data in
part because concepts are felt to be of greatest use to those developing
ceramic composites.
The more elaborate and topical introductions are given in the
beginning of each chapter.
CHAPTER II
SILICON CARBIDE FROM ORGANOSILANE PRECURSORS
Introduction
Many ceramic materials have specific properties that make them
ideal for energy related systems. Silicon carbide (SiC) is one of the
leading candidates for high temperature structural applications because
of its low density, high-temperature strength, chemical stability, re-
fractoriness, high thermal shock resistance, and creep resistance. To
achieve these desirable properties of silicon carbide, it is necessary
to develop a reproducible and reliable method for producing the material
in complex shapes and with a controlled ul trastructure.
In the conventional process for producing SiC material, silica in
the form of sand and carbon in the form of a coke are reacted together
at 2400°C in an electric furnace. The SiC produced is in relatively
large grains which are subsequently ground to the desired size.^^
The Cutler process^^ was developed to produce SiC material with
superior properties and cost effectiveness by using rice hulls. From
this process, the commercially known a-SiC whisker Silar" by AkCO is
obtained. The major advantage of it is that it has a much lower pro-
cessing temperature, ~1500°C, than the more traditional process.
The increasing search for new types of high-strength materials, and
for performance improvement in the existing ceramics, has pushed several
nonconventional approaches to ceramic synthesis.
10
11
As presented in Chapter I, obtaining nonoxide ceramics via pyroly-
sis of organometal 1 ic precursors has potential advantages over the con-
ventional methods of producing materials in low temperature process,
higher purity, fabrication of complex shapes, greater homogeneity, new
fabrication procedures leading to continuous fiber, coatings, and
impregnated porous structures. At a more fundamental level, polymer
routes can allow control over the microstructure of the intended ceramic
product with a unique combination of microstructure and phase assem-
blages and important consequences for both physical and chemical proper-
ties.'* Some of the more important applications of nonoxide ceramic
materials obtained via polymer routes are listed in Table II-l.
The first use of organic polymers to produce an inorganic refrac-
tory material was probably the development of graphite fiber from poly-
acrylonitri le in late 1950. ^'^ Other ceramic materials from organometal-
lic polymers v/ere first noted by Chantrell and Popper. 19 A partial
history of the development of nonoxide ceramics from organometal 1 ic pre-
cursors is given in Table II-2.'*
However, early workers50"52 ■\q organosi 1 anes (OS) genuinely be-
lieved that polysilanes were worthless and regarded them as undesirable
by-products of a faulty synthesis. Tnis all changed in 197b when Yajima
and his coworkers22-30 ,53-55 demonstrated that the polydimethyl si 1 ane
that was regarded as an undesirable by-product by previous investi-
gators50"52 is indeed a precursor to g-SiC. The reaction scheme, as
shown in Fig. II-l, is the polymerization of dimethyldichlorosil ane
[(CH3)2SiCl2] by dechlorination to yield polydimethyl silane (POMS). The
12
Table II-l.
Some of the Demonstrated Applications of
Nonoxide Cerauiics Derived fron Organonetal 1 ic Precursors
Form
Appl ications
Fiber
Reinforcement for composites
weaves, wovens, mattes
Mono! ith
Foam
Monolithic bodies for high temperature parts
Filters, packing, insulation, heat exchanger
Powder
Press to bulk body,
Fil ler material
Thin Film
High temperature electronic devices
Thermosetting polymer Metal -ceramic, ceramic-ceramic joints, bind-
er in powder forming
13
Table II-2
A Partial History of Monoxide Ceramics Via Polymer Psoutes
Precursor
Ceramic
Year
Polymers
Products
Investigators
Ref.
1960
phosphonitric
chlorides
P-N
Ainger, Herbert
18
1965
unknown
BN, AIM,
Si3N4, SiC
Chantrel 1 ,
Popper
19
1974-
■75
polysilanes
Si-C-N
Verbeek and
Winter et al .
20, 21
1976-
■81
polycarbosilanes
SiC
Yaj ima et al .
22-3U
1976
polyphenylborazole
BN
Taniguchi , Harada
Maeda
31
1978
carboranesiloxane
SiC-64C
Rice et al .
32, 33
1979-
■80
polycarbosi lanes
SiC
Scni 1 1 ing,
Williams, Wesson
34-39
1980-
■81
polysilastyrene
SiC
West et al.
40
polycarbosi lanes
SiC, Si-C-N
Baney and Gaul
41-46
1981
polytitanocarbosilane
Si-Ti-C
Yaj ima et al .
12
1982
polysilazanes
Si-C-N
Penn et al .
47
1982
polysilazanes
Si 3^4
Seyferth, Wiseman
48, 49
1984
vinylic polysilane
SiC
Schilling and
Wil 1 iams
39
14
c _
s
a
(/)
c
c
a.
E
r
9
C
o
nOi-6-i ^
I I I o
'^
<^?| — — '
ft x___x
c
o
E
T3
, C
o
o
E
o
Q.
EO
-O
>«
J3
C
o
c
-o
c
^3
CJ
-l->
c
u
fD
Z3
,—
T3
• r—
O
oo
s_
O
a.
-C
t.
CD
(O
-C
o
+->
>^
-o
o
c
c_
nD
ra
Ul
O)
>
+->
na
<_)
•r-
• »—
-o
oo
OJ
1
e
CQ
s_
OJ
QJ
-l->
U
c
3
*— '
■a
o
Ol
s_
-C
a.
4->
o
<u
+j
>
o
(/I
ja
ai
fO
OJ
u
e
o
3
1-
o
Q.
x:
oo
ro
cr
T3
• 1 —
r^
•f— ;
CJ
ra
■1—
>-
>-
o
z
en
s-
c
o
•r—
c
00
■ (—
c
r-^
OJ
1/1
4->
l/l
<J~>
o
s_
c:
(_)
o
•f—
c
■(->
Ol
o
Ol
(O
>1
<u
X
q:
o
I
o
o
w
o
lb
polymer chain then is rearranged to make it more reactive for a thermo-
setting condition. The rearranged polymer called polycarbosi 1 ane has
alternating silicon and carbon atoms. The thermosetting or cross! inking
carried out in air is a necessary step in the Yajima SiC fiber synthesis
to maintain the fiber shape during the subsequent heat treatments. Dur-
ing the heat treatment, hydrocarbon products are eliminated to yield a
ceramic char of SiC.
Since Yajima and coworkers22"30 »53-S5 developed a 6-SiC fiber with
excellent mechanical properties from the polycarbosi 1 ane, there have
been several other precursors potentially superior to the Yajima pro-
cess. 3'+"'-+° These are vinylic silanes developed by Wesson and
Wil 1 iams3'+"35 and Schilling et al.38.39 g.r\a polydimethyl phenylmethyl -
silane, better known as polysi 1 astyrene (PSS) developed by West et al.'+o
The vinylic silanes are reactive under a thermal crossl inking con-
dition but, because of their low viscosity (liquid at room temperature),
control of viscosity to draw fibers may require extra steps. On the
other hand, PSS is a solid with good solubility in common solvents and
possesses excellent tractability with good melt viscosity. However, it
possesses no reactive functional groups for crossl inking. It was noted
by West et al.'+o that the polymer strongly absorbs UV light at ~330 nm.
Irradiation of UV with x~330 nm on PSS was shown to crosslink the poly-
mer on the surface. '+0 However, the practicality of UV crossl inking of
PSS for larger structural ceramics is in question. An alternate way to
achieve bulk crosslinking is needed.
It is important to note that vinylic silanes and PSS are poten-
tially superior to polycarbosi lane because they can be processed without
16
the separate thermal rearrangement and oxygen crosslinking (Fig. II-l),
as required in the Yajima process. Oxygen crosslinking undoubtedly
introduces Si-O-Si in the network and ends up as silica in the final
product s-SiC. In order for the potential advantages of vinyl ic silanes
and PSS to be realized, it is essential that crosslinking methods be
developed which will avoid oxygen in tne SiC lattice following pyrol-
ysis.
The term polymer used here is defined as any organic or organo-
metallic compound that is not a monomer. However, in some specific
cases, oligomers are distinguished from polymers.
It is the objective of this work to investigate crosslinking
methods and some applications of vinyl ic silanes and PSS precursors to
SiC. In this chapter the following topics are investigated and dis-
cussed: 1) synthesis and characterization of the polymers, 2) modifica-
tion of the polymers for crosslinking, and 3) crosslinking and pyrolysis
to obtain SiC.
Experimental
Preparation of equimol ar dimethyl phenylmethyl copolymer
Reagent grade toluene for a solvent was dried with sodium metal in
~1 g sodium per 1 1 toluene by refluxing for 24 hours followed by dis-
tillation through a one way air sealed glass apparatus using mineral oil
bubbler.
A starting monomer dimethyldichlorosi lane {Me2SiCl2) from Aldrich
Chemical Co. was purified by distillation using a trap-vacuum technique
with liquid nitrogen. Methylphenyl dichlorosil ane (PhMeSiCl2) monomer
17
also from Aldrich Chemical Co. was vacuum distilled in a Yamato model
rotary evaporator at 80°C.
A reagent grade sodium metal bar was cut to 47.5 g and placed in a
dry 2 liter 3-necked round bottom flask. These operations were carried
out in a glove box with N2 atmosphere. The dried and distilled toluene
(850 ml) was added to the flask and the polymerization reaction appar-
atus was set up, as shown in Fig. II-2.
Sodium and toluene were mixed by stirring and heating to obtain a
molten mixture of sodium dispersed in the solvent. Heating was discon-
tinued to add dichlorosi lane monomers. A premixed solution of 61 ml of
MegSiCl?, 81 ml of PhMeSiCl2, and 50 ml of dry toluene was added slowly
through the air sealed side arm, while stirring was continued and N2 gas
was continuously flowing through the apparatus.
The rate of addition of the premixed dichlorosil ane monomers was
adjusted to maintain the gentle refluxing temperature of ~98°C, since
the initial dechlorination reaction is highly exothermic. A typical
duration of the monomer addition was ~30 min. Upon completion of the
monomer addition, external heating was restored to achieve a gentle
reflux. The reaction flask was kept dark by wrapping it with aluminum
foil. The reflux continued for 1.5 hours before cooling the reaction
mixture to room temperature and then poured into an isopropyl alcohol
bath (PrOH) with stirring.
Fractionation of the reaction products was carried out by first
separating them in PrOH. The polymer fraction was precipitated out
while the oligomer fraction remained in the solution. The oligomer
18
Silane
mixture
^2 out
variable
speed
motor
Thermometer
ToIuene-»- Sodium
+ Silane ^ v:.ir
Heating mantle
Fig. II-2. Apparatus for Polymerization Reactions for
Synthesizing Polysilastyrene
19
fraction in PrOH was collected by distilling off the solvent by a rotary
evaporator. The excess sodium residue was decomposed in PrUH. The oli-
gomer fraction collected was redissolved in toluene and washed with dis-
tilled water three times in a 200 ml separatory funnel to extract any
residual salt product. Then the toluene solution of the oligomer was
rotoevaporated to obtain a viscous oligomer fraction of PSS (PSS-0)
which was kept in a brown bottle after vacuum drying and weighing.
The PrOH insoluble fraction was washed with 200 ml PrOH and with
200 ml EtOH twice after draining the PrOH by filtering. After the solid
polymer fraction was dried in a vacuum oven for five hours at 55°C, it
was redissolved in warm toluene and the toluene insoluble fraction was
separated out. The toluene insoluble fraction was thought to be a
highly crosslinked polydimethyl si lane. This fraction was washed with
water five times in a separatory funnel and dried in a vacuum oven at
80°C for ten hours.
The toluene soluble fraction (PSS-P) in toluene solution was washed
with water five times to insure that all unreacted Si-Cl is hydrolyzed
out. This was done by titrating the effluent with AgN03 solution. Then
PSS-P in toluene was reprecipitated in 7 liters of PrOH. The bright
white precipitate was collected by filtration and dried in a vacuum oven
at ~50°C for ten hours. A pure PSS-P should appear as a white powder or
a clear, colorless solid.
The variations in reaction conditions for the subsequent runs are
given in Table II-3.
20
Table II-3. Summary of Reaction Conditions for
Preparation of Polysi 1 astyrene and Product Designation,
Mole Dichlorosilanes
Product
Mole Na
Addition
Reaction
Reaction
I.D.
Volume Toluene
Time, min.
Time, hr.
Temp., °C
PSS-10
0.5 mole each
PSS-IP
2.05 mole
1 1
PSS-20
0.5 mole each
PSS-2P
2.09 mole
1 1
A-PSS-P
0.5 mole MePhSiClg
A-PSS-0
0.4 mole Me2SiCl2
0.1 mole Allyl MeSiCl2
2.04 mole Na
1 1 toluene
4U 1.5 105
30 2 108
17 2 108
21
Infrared spectra for the polymerization products were taken by
using a Perkin-E1mer IR Spectrophotometer Model 283 v^ith KBr Pellet in
transmission mode and also by a Nicolet MX-1 FT-IR Spectrophotometer in
diffuse reflectance mode. Proton NMR spectra were obtained by using a
Varian XL-100 with CDCI3 or CgUg as solvents without the TMS reference.
Molecular weight distributions of the products were determined by gel
permeation chromatography (GPC) using polystyrene as a reference in THF
solvent and using a refractive index detector.
Other PSS samples were provided by Shinnisso Kako Co. of Japan
through the 3M Co. (J-PSSl and J-PSS2), courtesy of Dr. R. Sinclair and
also by Petrarch Systems, Inc. (P-PSS), courtesy of Dr. B. Arkles.
Vinyl ic silanes were provided by Union Carbide, courtesy of Dr. C.
Schilling. They are oligomer and polymer fractions of
Me3Si-f Si
■SiMe-
Two kinds of siloxane substituted PDMS containing a hydride functional
group or phenyl group were provided by Petrarch System, Inc., courtesy
of Dr. B. Arkles.
The structure of each polymer unit and physical state of all tne
polysilanes used in this study are given in Table II-4.
Crossl inking and pyrolysis
For crosslinking via y-f'^y irradiation, PSS was melt-coated on thin
stainless steel plates in glass test tubes with vacuum, Ar, M2, He, air.
22
Table II-4. Structure Forinulas of Polysilane Unit and
Physical State of the Organosi 1 anes at Room Temperature
Organosilane
Structure
Formul a
Physical State
At Room Temperature
PSS-P
-Me Me_
L-Ph Me-"
White to dul 1 yellow
sol id
PSS-0
,Me Me -,
li —Si--
ll-l
■-Ph Me-"
(where m < n)
Yel low to brownish
viscous liquid
Allylic PSS
A-PSS-P
A-PSS-0
Me
I
Si -Si-
I I
Me Ph
Light yel 1 ow sol id
Brownish viscous
1 i quid
Vinyl ic Si lanes
VI SP
ViSO
pMe^
MeoSi --Si
LI.
X L
rMe-,
Si
SiMe3 Colorless clear
viscous liquid
Y
Colorless clear low
viscosity 1 iquid
Siloxane PDMS
Hydride
-Si-O-Si- (SiMe2)8
Yel lowish viscous
clear 1 iquid
Phenyl
Ph
-Si-O-ii — (Si
(SiMe2)8
Yel lowish vi scous
cloudy 1 iquid
23
or NpO atmospheres. Sone portions of the silane precursors were dis-
solved in benzene in glass test tubes and sealed for irradiation. The
vinylic silanes were placed in evacuated Dorosilicate test tubes. The
glass tubes containing silane samples were irradiated witn y-radiation
from a *^'-'Co source at 1" distance for various lengths of time up to 29
days at room temperature.
Chemical free radical initiators (CFRI), benzoyl peroxide (BPO),
aszobi si sobutyronitril e (AIBN), and dicumyl peroxide (DCP) obtained from
Polyscience Co. were recrystal 1 ized from methanol before use. A few
grams of silane were dissolved in 5-10 ml of benzene in a test tube or
in a 3-neck round bottom flask and then the silane solution was degassed
with an inert gas. After 30-60 min., a CFRI in the range of 3-10 wt%
was added under an inert atmosphere and the crossl inking reaction was
carried out with heating on a hot plate or by a heating mantle, as shown
in Fig. II-3. The crosslinking reaction in the 3-neck flask was allowed
to reflux for twelve hours with continuous stirring before cooling to
room temperature. The crossl inked product was extracted and washed with
methanol .
Crosslinking via DCP was carried out also in a sealed Teflon con-
tainer for ViSP, ViSO, and PSS. About 1-2 g of vinylic silanes were
well mixed with 0.05-0.07 g DCP by a spatula under N2. Polysil astyrene
was also mixed with DCP after the polymer was made into a thick solution
in toluene. The silane + DCP mixtures were cured in an oven at 110-
150°C after the containers were tightly sealed. Other portions of poly-
sil anes were cured without DCP under the same condition.
24
N
2 out
Water
^out
Thermometer
acting
xture
Heating mantle
Magnetic stirrer
Fig. II-3. Apparatus for Crossl inking of Polysilastyrene
25
For crossl inking via Pf^"^, a 1.2 x 10"* M solution of Pt^"^ was
prepared by dissolving 3.1 mg of HoPtCl^'GHpO in 50 ml of a mixed sol-
vent of acetone (20 ml), ethyl ether (16 ml), toluene (10 ml), and EtOH
(4 ml). Crossl inking via Pt**"^ was tried for A-PSS, ViSP, and ViSO pre-
cursors by adding 10"^-10"^2 [^gles of Pf*"^ to 0.05-0.4 g of the silanes.
Crossl inkings of Petrarch's siloxane PDMS were carried out by
adding drops of concentrated THF/H2O solution of Zn(Ac)2, SnCl4,
[Co(en)3]2(S04)3, SnCl?, triethanol ami ne, ZrCl4, DCP, BPO, AIBN, Pt^"^,
and ethanolic NaOH followed by heating and curing in an air sealed glass
vial up to 15u°C or heating in N-^ gas up to 3U0°C.
Detection and confirmation of crossl inking of the polymers were
tested by using FT-IR, solubility in a solvent (benzene or THF), and
fusion at ~200°C. Pyrolysis of the polymers was carried out in a high
temperature furnace with an inert gas flowing at a rate of ~100 ml/min.
and also in a DuPont TGA 951 Thermogravimetric Analyzer with N2 or Ar
continuous flowing with a heating rate of lO'^C/min. Differential scan-
ning calorimetry (OSC) using the same DuPont Model 951 was carried out
in continuous Ar flowing with a heating rate of 5°C/min.
Crosslinking reaction products of PSS via DCP were identified by
GC's in order to elucidate the reaction mechanism. The product gasses
were introduced into a Tracor GC 550 and an HP 5880A GC. The detailed
experimental conditions and sample preparation are as follows:
Approximately 0.1 g of PSS-IP was dissolved in 1.5 ml of degassed
benzene, and then ~0.01 g DCP added and mixed in a 20 ml glass vial. The
thick solution was vacuum dried to evaporate benzene at room temperature
26
for tv/o hours. The dried sample (PSS/OCP) was placed in a ylass vial
with a rubber septun or in a Pyrex glass loop directly attacned to the
Tracor GC injection port. The samples in the glass containers were
heated to ~300°C by a Bunsen burner for 0.5-1 min. The product gasses
were either directly introduced to the silica gel column of a Tracor 550
through a valve or drawn by a syringe through the septum. The gas drawn
by a syringe was dissolved in a methylene chloride (MeCl2) solvent; a
few microliters of this solution was injected into the capillary column
of an HP 5880A GC
The GC parameters used dre given below.
Instrument: Tracor 550 and HP 5880A interfaced to an HP 85
computer
Column: Silica gel 2.7 m 60-200 mesh and glass capillary
Detector: Flame ionization
Column T: 45°C and programmed from 80°C to 250°C at 20°/
mi n .
Injector T: 180°C
Detector T: 180°C
Carrier Gas: N2
The overall experimental conditions for crossl inking of various OS
precursors are summarized in Table II-5.
Infrared spectra, SEM micrographs, and EDS spectra were obtained by
using a Nicolet MX-1 FT-IR Spectrophotometer and a JEOL model JSM-35C
electron microscope, respectively. The assignments of IR bands are
based on reference number 56 and are given in Table II-6.
27
a;
-a
IT3
CD
C
u—
• r—
j->
o
-i<;
c:
00
c
CJ
>^
JC
•^-
c
IT3
•!->
^—
+->
■a
CTl
en
fO
C
to
a;
a^
O)
O
s_
CM
_l
:_
H-
1
o
CO
iy>
£_
>i
-C
00
>1
O
TJ
V)
1
T3
=3-
1
to
13
c
'^
C
-ii
T3
•r—
1
• f—
0)
E
C_
c
0)
O
-C
3
CM
■a
c
o
to
c
to
to
>5
o
CD
c
•r— O
1/1
o
o
o
un
CM
I
o
o
o
o
en
ID
00
o
CO
— >
ai
to
c
=:
.1^
c
oo
*^~
00
(/)
in
O
►_
o
e
0)
o
t_
CD
■(->
-C
c
C2.
OJ
(/)
>
o
r—
^
o
+->
t/1
•<
0)
<D
0)
<D
0)
c
c:
c
c
C
0)
dJ
0)
OJ
OJ
•
Kl
•
ISI
3
Nl
N
•
t. O <J
c
o
c
C
C
(_>
r- !_ 0) CMTJ
i;
CMrO
(U
'o
<D
0)
13
13 «a: n: z >
-Q
-z.
>
J3
■U
J3
JD
>
OJ
c
• M
£- O C
•I- t_ IC O)
13 «=C > J3
e
(U
o
to
>,Lf5
13
c
<t3 •
4-)
fO
t- CM
(/)
<D
1 1
•^
2:
?-r-l
■o
>>
13
[_
13
c
>-
e
O)
>,
^
*-^^
i;_
en
fO
CL. O 03
Q.
OJ
>,
c_
O CL 1—
CJ>
-C
X
1
Q 22 <:
Q
■M
o
>-
0)
c
O.
13
1
r—
oo
r-
00
LO
Q.
o
I
28
Ul
in
</>
CO
>>
S-
>,
1-
lO
^
<T3
-o
-C
T3
o
^
-C
O
to
CM
CM
CM
t/i
(/> CO CO
(/I
— H
t- (/)
l/l
(/)
1
£_
C\J
t_ t_ s_
L.
1
x: t_
t-
C
C
-^
c
-C
1 — 1
x: -c -c
XI
cn
-E
x:
*^-
01
•r—
1
c_
ro
1=
E
<D
E
>*
LO
^ o 00
<a-
X!
1 C\J
CO
3
CVJ
.—1
OJ .^ <3-
CNJ
un ^H
<*
o
O
O
1
•
1 1 1
1
CO
• 1
1
t— (
1 — 1
n
evi
. — 1
o
i-H « — t 1 — 1
^H
r—*
O ^H
r-H
O
o
o
o
o
o
o
o
o o
O
O
IX)
CO
CO
p^
O Lf)
O
Lf)
:\j
o
1
o
r—f
1
CO r-(
1 O 1
CO
1
O
■-H
^
1-n
o
o
o
o
O C\J LD
o
1—
^H LT)
O
CM
.— <
CO
I— t
I~~
r^
r-~ ^H i^
r~
□£
.— t :^^
r--
T3
e
o
o
LO
1
(U
01
0)
rsi
'~~
C Ll_ t-
t_
XJ
O) 3: CM--
OO
•^
03
xj h- z: Tj
ns
"O
T3
-o
■a
T3
a;
1)
a;
(U
<U
^_
^—
^*
r—
^—
O) IT3
(O
«3
fO
13
C (U
a;
q;
0)
(U
aj I/)
00
CO
to
CO
3
•
•
•
— £_
;_
L.
t.
O
O
c_
O
O OOt-
.f—
• r—
•r—
ro
ro
•r—
to
+-> :^ Tj
fO
TJ
-fl
>
>
ro
>
CLOcQ ai>i O) Cl.«^ dJ Q-'^ (U i- Q. O CO ':r
(_3Q._ .EX .C 04-> JTO-I-' .CICJ<:i-^--l->
o. to to
I •!- -r-
«c > >■
29
T3
o
O
o
o
CO
I
o
■t-)
c
o
o
— ' <U
(/)
Ol
' — t_
-CI CSJ -r-
"3 S: 03
CO
o o
00 ns
— - Z
o
CM
evj
1 — 1 O
3:
C^i-
ir>
CSi-^.
•
c o
'^ >* C\JO
QJ C
r— r— r—
<c
-^ rT3
2= +
CJ> o o
O ^
Q.
O
ccw
c c c
c
O 4-)
<_)
a.
1— 1 +
rvi </1 oo
rvi 1 — 10)
Q
en
< Q
o
30
Table II-6. Infrared Absorption dands of Polysilanes
V. cm" Mode Shape, Intensity
300 Si -Si bending ' weak
420 H-Ph rocking weak
400-480 Si-Si stretching
700-800 Si-C stretching broad
800-850 Si-CH3 rocking
850-1000 Phenyl C-H bending
1020 Si-CH2-Si wagging shoulder
1100 Si-O-Si stretching sharp
Si-Ph medium
1250 Si-CH3 bending strong
1400 -CH3 deformation broad, strong
1480-1580 Aromatic C=C weak, sharp
1620 OH bending of adsorbed water
1600-1680 C=C aliphatic
1710 C=0 medium
1800 -Ph broad
2100 Si-H sharp, medium
2900 C-H stretching in Si-CH3 strong
3050 C-H stretching in atomatic narrow
3450 OH stretching broad
3630 Si-OH stretching broad
31
Results
Characterizations of the precursor silanes
Some of typical IR, EDS, and NMR spectra are given in Fiys. II-4
through 11-19. The M. W. distributions, X yield of each fraction, and
MePhSi/Me2Si ratio for PSS are given in Table 1 1-7. The MePhSi/Me^Si
ratios were estimated by integrating the deed under the peaks and nor-
malized by the number of protons in each group of the peak.
The oligomer fraction of PSS-1 in Fig. II-4 shows some C-OH and
Si-OH (-3300 cm"^ and 3600 cm"M, Si-H (-2100 cm'M, possibly some un-
saturated carbon, i.e. C=C (-1600-1900 cm"-^), strong and sharp Si-Me
stretch (-1250 cm"-'-), the strong and broad band for Si-O-Si, and Si-Ph
overlapped with Si-O-Si (-1100 cm"M, -Ph (-700 cm"^), and an Si-Si
stretching band at -450 cm" . In PSS-IP, there is not as much Si-OH and
little C-OH, less Si-H, and a small but sharp peak for Si-Ph at -1100
cm"-'- is shown (Fig. II-5).
Figure II-6 suggests that the oligomer fraction of PSS-1 has a more
complex structure, indicated by multplets of the CH3 region (-1-2 ppm 5
scale) and the Ph-H region (8-9 ppm), than the corresponding structure
of the polymer fraction. It also shows a possible C=C band at -4.9 ppm
which is absent in the polymer fraction (Fig. II-7). The peak at -2.6
ppm may be ascribed to to -CH2- or to Si-H.
Figure II-8 for PSS-20 shows that a larger proportion of Si-H is
present in the oligomer, but less Si-OH and C-OH is present than in
PSS-10 (Fig. II-6). Although PSS-10 and PSS-20 are both oligomer
fractions of PSS, the IR spectra show that they are not exactly the same
32
>,
>i
O
o
u
o
en
O
03
3
S_
o
Ol
Q.
-o
S-
(T3
o
o
o o
00 ID
30NVllllNSNVdl %
o
o
(M
O^
33
o
<^
CJ
o
o
o
o
o
o
^^
GO
Q.
o
1
o
Q
Q-
o
Ol
o
c
>>
O
U1
fT3
o
^*
^
e
(/I
u
>>
o
o
o
CO
Q-
{£
oc
M-
LU
O
GQ
C
o
o
2
3
o
®
Z
UJ
s_
o
>
u.
o
o
%
O)
E
CJ
o
o
<^
o
in
o
CJ
E
i.
o
o
o
t/1
ro
O
O
<4-
C
1—4
ro
ID
8
1
1—4
»— 1
^^
•
o
'^
■1—
30NVlllWSNVyi %
34
en
o
b
o
o
GO
on
a.
^
3
U
1
""■
a.
o^
aj
in -'
2
rrj
Q_
o
Q.
LO
o
O)
I
oi
o
ra
1^
s_
ra
<u
r^
o
3
o
nn
c
o
-t-J
O
s_
c
O
a>
1
35
36
o
I
Cl.
13
S_
*->
u
QJ
Q.
OO
■a
o
s_
tJ
CO
30NVlli;NSNVdl 7o
37
OJ
E
O
CD
c
4-
O
O)
Q.
^>^
■a
I
33NVllllNSNVdl7o
38
O
O
O
O
ro
O
O
00
o
o
C71
U~>
Cl.
>i
o
E
3
"sT
u
HI
^-^
Q.
CO
(/I
o^
a;
(J
"^lE
Q_
o
Q-
to
oe
O
o
•f—
c
<J
3
C
O
o
39
O
O
o
o
CM
O
o
CD
O
O
CD
O
O
CD
o
o
o
o
o
CJ
o
o
^ T^
s_
- E
03
>1
,«
'0
CD ^
Q.
— cr
00
00
CD
Q_
92
(_)
00^
z
'>,
LjJ
'
^
5
o<
CO >
4-
E
3
s_
4_)
10
GJ
Q.
CM
l/^
TD
QJ
ra
s
>+-
■— '
•
-—
10
1
ro
•— t
C7^
Ll.
30NVllll^SNVdl %
40
O
d
o
CM
o
ro
o
1.
^
E
o
Q.
o
**^
Ul
iri
^
2
""
a.
>!
CL
■
o
<
ID
tt-
o
O
00
o
cjS
o
d
13
S_
■4->
U
OJ
CL
c
o
■»-)
o
i.
Q.
41
O
d
u
I
o
cvj
O
rO
O
i.
E
o
C-
D-
o
Q.
O
1^
O
00
O
O
d
Ol
a.
o
42
to
oo
3
S_
+-)
U
OJ
Q-
>
s_
O)
o.
to
>1
I
s;unoo
43
Q.
13
o
l/l
c
c
>
o
B
3
5.
■I-J
U
Ol
Q.
1/5
I
44
O
o
CVJ
O
O
CO
o
o
^
^
•
£
o
^-^
O cn
OQ^
O^
cvjOQ
:^
ID
Z
liJ
oo
>
oo
1
(M
<t-
fO
E
3
i.
+J
o
OJ
o.
oo
o
c:
o
^
1—
^
U-
t'IZQ
Z6Str
08i7£
£9£Z
9trZ
33NV103nd3d %
o
o
in
45
o
o
c\J
o
t/1
O
O
=>
CD
<tJ
^—
•r-
OO
U
O
^
?r
- E
>
o
4-
^■^
O
CO
a:
o
OLiJ
OQQ
u
OS
03
CMZ)
L^
z
S_
UJ
OJ
5
o
C7)
8^
CM
o
O
3
S-
U
<u
562171 06601
98SZ 18 It^
30NV133nd3a %
o
O
46
- O
- C\J
- rO
i-
<u
E
'd-
O
•^
1—
o
O)
^""^
c
in^
TJ
£
(yO
Q.
U
O.
- CD
— r-
— CO
- 0>
- o
3
S-
-4-)
O
<u
a.
oo
q:
o
S-
a.
CO
I
47
V
o
oi
c
o
c
3
S-
■t->
o
(U
o.
CO
o
■t->
o
I
48
Table II-7. M. W. Distribution, % Yield,
and the Ratio of MePhSi/Me2Si for PSS
%
Yield
Ratio
M. W.
(Mn)
01
iyomer
MePhSi/Me2Si
01 igomer
Polymer
01 igomer
Silane
Polymer
Insol
. Polymer
72
Polymer
PSS-10
390
1.05
103
19
1.1
PSS-IP
10 X
8
PSS-20
344
T
72
15
1.20
1.35
PSS-2P
11 X
10^
11
A-PSS-P
331
o
75
12
1.1
1.2
A-PSS-0
11 X
103
11
J-PSSl
6.8 X 10^
49
compound; see qualitative differences at ~1720 cm" , ~1580 cm"-'-, ~1250
cm"-^ , and -450 cm" .
Figures II-9 and 11-10 show that the allylic PSS oligomer has a
greater unsaturated carbon group, as indicated by the broader band at
~16U0 cm"-'-; this is supported by the NMR data which show small humps
around 3-7 ppm. However, this cannot be conclusive because the bending
mode of water is also at ~1600 cm" . Tne band shape at ~145U cm"-*- is
different from the nonallylic PSS. A significant amount of Si-O-Si may
be present in A-PSS-0. Groups including possibly Si-H and an allyl
group represent ~4% of the total protons based on the NMR data. Allylic
PSS polymer represented by Figs. 11-11 and 11-12 contains a smaller pro-
portion of Si-H relative to C=C. Carbon-13 NMR (Fig. 11-13) does not
reveal any additional information.
Figure 11-14 shows that only Si as a metallic element is present in
PSS-IP under EDS analysis.
Figure 11-15 for the solvent insoluble fraction of PSS-1 presumably
due to the crossl inked network indicates that the polymer may be mainly
composed of
,Me Me
i— a.
Me Me^"
An infrared spectrum of J-PSSl (Fig. 11-16) obtained from the diffuse
reflectance mode of FT-IR indicates that Si-H is present, as well as
-OH, but not as much Si-O-Si is shown. An IR spectrum for ViSO (Fig.
11-17) obtained by the same way as for J-PSSl shows a larger proportion
of Si-H. A small but sharp band at ~1600 cm~^ may be that of C=C. The
50
NMR spectrun of ViSO in Fig. 11-18 shows the presence of a vinyl group
at 6.2 ppn. It is difficult to see the presence of Si-H, although snail
humps between 1-4 ppm are shown. The NMR spectrun for ViSP (Fig. 11-19)
indicates that the polymer has essentially the same structure as the
ol igomer ViSO.
Crossl inking and pyrolysis
Si lanes exposed to a y-ray dose greater than 200 Mrad (~30 days at
1" distance from the source at 0.3 Mrad/hr) were infusible at tempera-
tures above 200°C and were insoluble in THF or in benzene, indicative of
crossl inking. Additions of CFRI prior to y-ray irradiation made no
difference in the crosslinking reaction rate.
Fourier Transform IR spectra of PSS samples before and after 29
days of y-ray irradiation are shown in Fig. 11-20. The polymer film on
a stainless steel plate after 29 days of irradiation in vacuum showed
insolubility in THF and infusibility upon heating up to 250°C.
Gamma-ray irradiation of the vinyl silane oligomer in vacuum showed
an increase in viscosity within twelve days from a watery fluid to a
semisolid form. The vinyl silane upon heating at ~200°C for 10 min in
N2 was transformed into a light yellow translucent solid which was in-
soluble in toluene.
Gamma-ray irradiation of PSS in a NoO atmosphere for >11 days
changed the color of PSS from translucent yellow-green to bright red-
brown. The viscosity of PSS decreased sharply, indicating the degrada-
tion of the polymer.
Among the CFRI's investigated (BPO, AIBN, and DCP), only DCP in the
range of 2-15 wt% in PSS/benzene solution showed a positive crosslinking
51
2000
1466 1199 932
WAVENUMBERSlcm-')
398
Fiq. 11-20. Reflectance FT-IR Spectra of PSS Before and After
29 Days of V-ray Irradiation in Vacuum
52
reaction. The CFRI DCP has the highest decomposition temperature of
~150°C of all other CFRIs. The FT-IR spectrum of DCP crossl inked PSS at
250°C compared with that of the as-synthesized PSS is shown in Fig. II-
21. The DCP reacted polymers were insoluble in benzene and infusible at
temperatures above 20U°C. The density of PSS crossl inked by DCP measur-
ed by mercury volume displacement was 0.77 g/ml . Fourier Transform IK
spectra of A-PSS samples after they were treated at different crossl ink-
ing conditions are compared in Figs. 11-22 and 11-23. Oxygen cross-
linking by heating in the air at 25-80°C came out negative for all PSS
precursors.
In the crossl inking of ViSP without DCP under the same conditions
as with DCP, no solidification was observed within 20 hrs. However,
curing at 30°C higher temperature 140°C resulted in solidification of
the liquid ViSP, signifying crossl inking.
The difference in the chemical structure of the ViSP samples cross-
linked thermally compared with ViSP samples crosslinked with DCP is
shown by FT-IR spectra in Fig. 11-24.
Chromatograms of the gaseous products from a crossl inking reaction
of PSS with DCP after being separated by a GC is given in Figs. II-2b
and 11-26. Approximately 100 times more methane is produced as compared
with ethane, as shown in Fig. 11-25.
Differential scanning calorimetry thermograms are given in Figs.
11-27 and 11-28 to compare the crosslinking mechanisms. In Fig. 11-29,
DSCs of oligomer and polymer PSS are compared. Scanning electron micro-
graphs of the surface of DCP crosslinked and pyrolyzed PSS and ViSP are
shown in Figs. 11-30 and 11-31.
53
2000 1733 1466 1199 932
WAVENUMBERS (cm-')
665 398
Fig. 11-21. FT-IR Spectra of PSS-1 and PSS-1 Crosslinked
With 8% DCP
54
O
o
o
o
CD
00 c
1:5
o
o
Si
en
o
T3
c
ra
<u
i-
o
>,
—)
C\J
QJ
Q.
00
CM
CM
I
80UDp9|^9y%
55
on
>,
o
o
y%
Id
c
U J
o
+-'
01
>i
.—
ra
-»->
s
^_
r-
1
-r
E
■a-
o
D-
^
l*-
o
Qd
o
-1-J
o
III
o
DQ
S
UJ
3
Z
•^
Ld
c
o
>
CO
i
o
s_
O
(U
■!->
O
o
l£2
5
o
on
«
i-
+J
o
o
in
Q.
to
^
c:
1
1—
u.
o
O
1
Q
»-H
OvJ
•r—
U-
56
ViSP/DCP,110 C,
UJ
o
z
<
o
UJ
_l
u.
u
5600 4400 3200 2000 1400
WAVENUMBERS (cm-*)
800
Fig. 11-24. FT-IR Spectra of Crosslinked Vinylic Silane
Showing the Effect of DCP, Temp., and Time.
57
CO _
O
Q
if)
£Oi;cozi
J
1 1
a:
-.
—
-fO
CO Z
-O
u
D-
3
<_)
■0
^^
S-
OO
Q-
t/0
C^
s_
e
■<—)
(C
S-
21
-a
O)
c
i.
t3
D-
Ji:
■(->
ly)
Ol
3
2L
O)
cn
ai
e
t3
• f—
C3
3
a
JZ
-C
oo
■I-'
c
>4-
•1—
+J
CO
u
E
fO
ro
O)
S_
Oi
cr.
O)
4->
c
fO
•r—
f=
-^
c
i.
•f—
-C
r—
C_)
1/1
l/>
to
to
s_
C5
(_)
Ln
CM
Lj..
58
tR, MIN
B
Fig. 11-26. Gas Chromatograms of the Reference and the Reaction
Product From PSS/DCP Crossl inking Reaction Dissolved
in Methylene Chloride (MeCl2). The Numbers on the
Peaks Are the Retention Times in Minute.
59
V
c
Ui
I
O
i-H
o
X
Ul
PSS- I P
no OCP
ex situ
100
200
300
400
500
600
Temp. °C
Fig. 11-27. DSC Thermograms of PSSl and PSS1/DCP Before and
After the Crossl inking Reaction
60
■0
I
n
u
600
Temp. ^ °C
Fig. 11-28. DSC Thermograms of J-PSS2 and J-PSS2/DCP After the
Crosl inking Reaction
61
a
o
r~>
--^
PSS-IO/DCP
J\
\
^ Ciosslinking ,
\ ■
I \ in situ
y
\
r
100
200
300
400
500
600
Temp.^ °C
Fig. 11-29. DSC Thermograms of PSS-1 Oligomer and PSS-1 Polymer
Reacting in situ With DCP
62
Fig. 11-30. SEM Micrographs of PSS/DCP Showinx Blisters and Pores
Generated by Gas Evolution. Top: After Crossl inking and
Pyrolysis at 400°C in Vacuum, Bottom: After Crossl inking
and Pyrolysis at 900°C in Nitrogen
53
Fig. 11-31. SEM Micrographs of ViSP/DCP Showing Pores and Surface
Texture. Top: After ViSP Cross! inked With DCP at nO°C,
Bottom: After Crossl inking and Pyrolysis at 900°C in N
2
64
The more effective cross! inking conditions among the techniques
tested on the various silanes are summarized in Table II-8.
TGA thermograms of the precursor silanes are given in Figs. 11-32
through 11-37 to show the char yield of SiC. Fourier Transform IR spec-
tra of the pyrolyzed products are given in Figs. 11-38 through 11-42.
In Fig. 11-39, the SiC product from PSS/DCP is compared with the commer-
cial e-SiC Nicalon**. The spectra show the characteristic absorption
band of Si-C stretching at 793 cm"! along with a small SiO^ band at
-1040 cm"-^. An XRD powder pattern of PSS showed that the pyrolyzed
product is amorphous which is identical with Nicalon®.
The char yield of PSS without crossl inking was less than 20 wt%,
which is close to the char yield of ViSO, while the char yield of the
PSS/DCP systems show 52-61 wt% SiC. The char yields of pyrolyzed prod-
ucts of SiC from various silane precursors are listed in Table II-9. An
XPS spectrum for Si2P of SiC derived from ViSP in Fig. 11-43 shows -20
atom% oxide silicon on the surface of SiC indicated by an overlapped
peak at -107 eV B.E.
Di scussion
Based on IR and NMR data, PSS-IP contains a low level of Si-OH
(-3400 cm"l) and Si-H (-2100 cm"^) bonds. Unsaturated carbon components
(-1600 cm"-'-) may also be present at a low level. In PSS-10, significant
amounts of Si-O-Si overlapped with Si-Ph, as shown by the broad absorp-
tion band at -1100 cm" . The formation of Si-O-Si may be caused by
water used to hydrolyze the residual Si-Cl in the polymer and the alco-
hol solvent used for fractionation.
65
Table II-8. Summary of Effective Crosslinking Conditions
Found for Different Si lane Systems
Sil ane
PSS-P
Means
Y-ray
CFRI
Effective Conditions
vac. 29 days at 2.56 cm
RT
DCP 110-200°C
in 10 min-10 hrs
absence of oxyyen
PSS-0
CFRI
DCP, 140-2bO°C
in 2U min-12 hrs
absence of oxygen
A-PSS
Thermal
CFRI
Pt4+
300°C in 20 min
or 170°C in 14 hrs
absence of oxygen
IbO^C, 20 hrs
absence of oxygen
80°C, 12 hrs
ViSP
ViSO
Thermal
CFRI
Thermal
Y-ray
CFRI
150°C, 24 hrs
absence of oxygen
DCP, 120°C, 4 hrs
absence of oxygen
> 200°C, 72 hrs
10 days, polymerization
not a cross! inking
DCP/110°C, ~3 hrs
or 75°C, -12 hrs
no oxyyen
Ph
SiOSiMeg
CFRI
DCP/N2, 300°C
20 min
66
200
400 600
TEMP. °C
800
1000
Fig. 11-32. TGA Char Yields of PSS-10, PSS-10 After DCP
Cross! inking, and ViSO
57
100 1
80
UJ
9 60
CO
111
a:
^40
H
X
O
LlI
^ 20
ex situ
PSS/DCP 7%
PSS
ALLYL PSS
_L
200
400
TEMP
600
800
1000
Fig. 11-33. TGA Char Yields of SiC From P5S-1P, PSS-IP
After DCP Crossl inking, and Allylic PSS
68
CD
>-
S-
03
3
O
00
(J
S_
X2
3
(U
>
o
I
3nais3y
3,n
69
80
§ 60
CO
u
J ^0 1.
20 -
\ \
\ V
-
\ " " - _ _ ViSP/6% DCP
^
^^ ViSO/6% DC?
1 1 1 1
200
AOO
600
TEMP.
800
1000
1200
Fig. 11-35. TGA Char Yields of SiC for ViSP and ViSO After
Cross! inked With DCP
70
o
i.
o
-o
c
OJ
s_
o
OJ
ca
Q.
I
c
i.
o
in
c
c
T3
>-
O
C
Char
DCP
sz
O
c
•a: +->
O -1-
vo
1
Ol
71
en
M
u
U
fsi
^'
."^"
o
O
in
U-.
CNl
3
■ —
eu
4-1
•k
u
•H
&.
c
o;
U
Q
o
1—1
c
a.
o
CL
CJ
CNl
1 — I
1
a
in
r/1
cr.
oNi
a.
m
o
1
a.
.—
—1
o
c
CO
c
c
t- -J
c a.
c
c
-fc-
o
4-
(U
en o
O ro
00 Ci
.
CT>
c/0
C
oo
• r—
Q-
-i^
d
s_
*^-
o
1
4-
CO
l>1
o
O
■1 —
s-
</1
o
^
O)
o
-C
+->
(/>
T3
M-
^
O
0)
a;
>-
s_
3
s_
+->
ro
n3
-C
S-
<_)
0)
Q.
•=X
e
o
0)
h-
1—
t-
I
c
00
o
v£3
o
anoisaa % -^m
72
(13
1/5
00
>^
o
i-
O-
s_
ZZ2B 6606 0Z8-8 It^Q
30NVlD3nd3d %
Z92B^
C
XI
•r- OJ
.— s_
to OJ
1/1 -C
O Q.
S_ </l
o o
E
C_ -(J
O CD
o
e s-
:3 4->
s_ •.-
■i-j 2:
u
OJ c
Q.---
»— »
I O
h- U3
u. cn
00
I
73
o
1
Q_
-
n
-
■^
00
>.
J>
oo
^
^ - ~~^
^^
CL
1 1 '
<
o
_J
"^'^
1 1
-1.^
^
1
1
/
\ \
\ 1 _
//
\
\
\
\
1 \
o
o
o
o
O
o
CO
o
o
o
o
o
£
o
LU
00
LU
08|-9 ^889 i6Qg Z6Zg 2009^
c
o
o
•r—
c
•r—
®_
o
U
"3
I
CO
•r-
74
o
o
o
o
CM
OO
C-
I
'-J
3
S_
u
c
DC
o
1
62t7-b
6t^£
t792£
6Z9 2 t'SOZlS
30NV103nd3a 7o
JO
75
O
O
O
O
CO
o
O
o
O
^
«-«
cn
^
-t-"
£
ro
o
1/1
'^-'
C/D
t/1
o
o
o
(M
a:
m
o
s-
2
D-
i-
0)
-l-J
llJ
03
>
o
<
D-
o
:$■
Q
ro
00
S99
S2> S0£
30NV103nd3d %
o
o
o
o
in
3
S-
+->
O
Q.
OO
I
76
-C -r-
O-
-c c:
OJ -r-
s-
E fO
O J3
C
C_ I
C 4-
OT O
r> O)
u
E c
O O)
>+- j::
■o
•r- 4->
ro
+J O)
c o
O n3
E *'~
S_
-l-J y-
u o
o
Q- E
S-
c: +->
<—• o
I OJ
I— CL
Li. 1/1
I
y %
CD
•r—
Sil ane
PSS-IP
PSS-2P
J-PSSl
J-PSS2
P-PSS
A-PSS-P
ViSO
PSS-10
77
Table II-9. Char Yield of Pyrolyzed Products SiC
froin Various Si lane Precursors
Crossl inking method Pyrolysis Char Yield
and Conditions Method %
10% DCP in situ TGA 30
10% DCP, 250°C TGA 61
20 min, vacuum
7% DCP, 250°C TGA 52
10 min, N2
5% DCP, 110°C TGA 23
10 hrs, air sealed
8% DCP, 250°C TGA 52
20 min, N2
- TGA 12
800°C, 1 hr, N2 2
5% DCP, 150°C, 12 hrs TGA 25
air sealed
5% DCP, 130°C, 12 hrs TGA 20
air sealed
TGA 15
TGA 21
Pt^^, 80°C, 12 hrs TGA 35
5% DCP, 170°C 900°C, 1 hr. 22
TGA 25
4% DCP, 110°C, 6 hrs, TGA 25
air sealed
8% DCP, 150°C, 12 hrs, TGA 16
air sealed
78
Table II-9 (continued).
Sil ane
PSS-20
Crossl inking method
and Conditions
7% DCP, 300°C
2U min, N2
Pyrolysi s
Method
TGA
Char Yield
%
67
Insol . PSS
TGA
Insol. A-PSS
TGA
ViSP
thermal , 150°C, 24 hrs
3% DCP, 120°C, 3 hrs
6% DCP, 130°C, 4 hrs
920°C, 1 hr, iN^ 55
920°C, 1 hr, N2 54
TGA 72
Si-0Si-Me2
TGA
<4
Ph
I
Si-0Si-Me2
8% DCP, 250°C
TGA
27
79
o
o
siNnoo
m
I
80
All of NMR spectra show slight excess of the PhSiMe unit in the PSS
copolymer chain over the MeSiMe unit, despite the intention to form an
equimolar copolymer. This must mean that the PhSiMe monomer unit is
more reactive than the MeSiMe unit during the polymerization reaction.
This means that in order to achieve an exactly equimolar copolymer of
PSS, one would have to use a slight excess of Me^SiClj monomer.
Hydrogen directly bonded to silicon in PSS comes from the fact that
the polysilane chain ends are probably anionic in the sodium/toluene
milieu and abstract hydrogen from the alcohol that was added to quench
excess sodium. Since the Si-H is hydrolyzable to give Si-UH, when tne
Si-H ends find Si-OH ends, they form a Si-O-Si linkage by a condensation
reaction. „
A
Although early workers of polysi 1 ane5°"52 considered that 4-Si4- from
VJ '"
Me
MeSiCl2 was useless because of its insolubility, introduction of a
phenyl group improves its solubility and tractabil ity.
As given in Table II-6, all polymerization reactions carried out
produced ~3 times larger oligomer fractions than polymer fractions.
Although it is generally thought that polymer fractions are tne desired
product in these reactions, tne oligomer was successfully repolymerized
and crossl inked by using a DCP as a CFRI. This is an especially import-
ant finding since the oligomer fraction in a liquid state at room temp-
erature is more convenient to impregnate porous ceramic bodies in order
to strengthen them. David, ^6 who pioneered PSS, has unsuccessfully
tried to repolymerize the oligomer fraction by restarting the polymeriz-
ation reaction with sodium in toluene, sodium chloride, lithium metal,
lithium t-butoxide, potassium t-butoxide, or sodium with biphenyl as an
81
electron transfer agent. Ddvid56 clearly concluded that "all one can do
is to fractionate them out of a polysi 1 astyrene product and discard
them, . . ." Petrarch System, Inc., the only company that makes PSS
commercially in the U.S., follows this practice. 57
Oligomers of PSS are believed to be in cyclic form so that the
crossl inking mechanism is expected to be different from the polymer
fraction by opening up the six membered ring. This is partly evidenced
by a higher TGA yield (67%) than that of a polymer (Figs. 11-32 and
11-33). The DSC data (Fig. 11-29) show that oligomers require a higher
temperature for crossl inking than polymers and the decomposition begins
at ~50°C lower temperature than polymers. The high char yield of the
oligomers, however, was partly caused by the loss of volatiles through
the vacuum line during the ex situ crosslinking reaction with UCP.
The longer reaction time of PSS polymerization, e.g. ID hrs, was
believed to degrade the formed polymer by excess hot sodium. However,
it appears that a longer reaction time than ~2 hours could have been
used to increase the polymer fraction of PSS. It is not yet completely
clear to what extent the longer reaction time improves the yield of the
polymer fraction. A systematic study of the effect of reaction time on
the M. W. of the silane products must be continued.
In the allylic PSS polymer, small amounts of Si-O-Si and Si-H are
shown (Fig. 11-11). The unsaturated carbon component, probably from the
allyl group, is shown at ~1600 en"-'-. The mole ratio of MeSiPh to MeSiMe
is 1.2:1. The i^C NMR (Fig. 11-13) did not reveal any more evidence of
the presence of an allyl group in A-PSS-P. Some indications of allyl
groups in A-PSS-0 are also shown in Fig. II-9. The group representing
82
Si-H and the ally! is shown as -4% of the total protons in A-PSS-P (Fig.
11-12). Apparently, not all the allyl methyl dichl orosi 1 ane ended up in
the products. A further study to account for this is needed.
The low solubility of J-PSSl, J-PSS2, and P-PSS in the solvents
must be a result of incomplete fractionation of the insoluble high M. W.
fraction. This variation in the M. W. distribution in a polymer is most
likely affected by the fractionation procedures. The consequence of
this was observed in the different behavior of PSS under the same cross-
linking conditions used in this study.
The average number M. W. (Mn) for different PSS precursors are in
close agreement, except that of J-PSSl. The GPC chromatograms show that
the PSS polymer is bimodal with a ~4 times larger lower M. W.
portion (Mn 9 x lO^) than the higher M. W. (Mn 3 xlO^).
The very high M. W. fraction that is insoluble in common solvents
and infusible upon heating, appears to be polydimethyl sil ane with some
Si-H, Si-Ph, and Si-O-Si, as shown by the IR spectrum in Fig. 11-15.
The 1 inear
Me
linkage apparently has been crossl inked via bridging oxygens of Si-O-Si
type. The TGA char yield of this fraction was <10% (Fig. 11-34), which
agrees with the result of Yajima et aU^"*
In vinylic silanes, a large and sharp Si-H band at ~2080 cm"-^ is
shown for ViSO in Fig. 11-16. The vinyl double bond appeared at ~1650
cm"-^. However, the hydride proton is not apparent in NMR (Fig. 11-18).
83
This should be due to the relatively low concentration of the hydride,
but this obviously disagrees with the IR data. This point will be
discussed further in the latter part of this section.
An energy equivalent to 200 Mrad or greater y-ray irradiation
required to crosslink PSS is not unusual because of the phenyl group^s
on the chain and the absence of an Si-H functional group.
In Fig. 11-20 the effect of y-ray irradiation on the PSS structure
after 29 days is shown and the sharp bands at ~70U cin"-^ representing the
methyl group are lost. The DCP reacted PSS lost most of its IR bands
for Si-CHj, and Si-H (Fig. 11-21).
Tne enhanced crossl inking of carbon polymers in NpO under y-irradi-
ation, as shown by 0kada,59 did not occur with silanes but rather the
opposite was observed.
Among the several CFRI studied, only DCP yielded an insoluble and
infusible solid of polysilanes. This is probably due to the active
methyl radical, which was not present in any other CFRI used. In the
crossl inking reaction, DC? has to be decomposed to give the methyl
radicals. This occurs at ~150-2G0°C.
Allylic PSS is observed to be more reactive under crossl inking con-
ditions than PSS. As shown in Fig. 11-22, A-PSS can be crossl inked
thermally or via use of the CFRI DCP. For thermal crossl inking, a temp-
erature >170°C is required for complete reaction. Allylic PSS is also
shown to be crossl inkabl e by Pt'^''" catalyst. At least 2.4 x 10"^ mole
Pt'^ per ~0.3 g A-PSS was required for an effective crossl inking of
A-PSS, as shown by IR spectra (Fig. 11-23).
84
The crossl inKing reaction of A-PSS must be between Si-H and C = C,
as shown in equation II-l.
catalyst
= SiH + C=C > = Si-C-C-H (II-l)
The coupling reaction is catalyzed by Pt^"*".60 jhe TGA yield of A-PSiJ
without a precrossl inking treatment (21%) is still greater than that of
PSS-P (12%). After being crossl inked with Pt"^"^, the yield increased to
35%. Differential scanning calorimetry (Fig. 11-44) shows that the
crosslinking between Si-H and \= occurs at ~240°C. The CFRI DCP re-
quires temperatures >170°C for complete crosslinking, but Pt"^"*" catalyzed
the reaction at a temperature of ~80°C. This crosslinking reaction by
Pt^"*" is unique to A-PSS and demonstrates the advantage of incorporating
an allyl group into polysilane synthesis.
Although a monomer with an Si-H functional group was not added in
the A-PSS synthesis, the small amount of Si-H at the chain ends was
still shown to be effective in the coupling reaction. However, inten-
tional small amounts of a monomer with a Si-H functional group, e.g.
H
CI Si — Me should improve the crossl inkabil ity even further.
2
Vinylic silanes can be crosslinked both thermally and via the CFRI
DCP. With DCP, the crosslinking is achieved faster and required a lower
temperature: 110°C for ~4 hours as compared to 150°C for 12 hours with-
out DCP. Without DCP, 150°C for 12 hours treatment still did not pro-
duce complete crosslinking, as shown by the large Si-H IR peak in Fig.
11-24. This is also shown in Fig. 11-45 with an expanded scale.
The vinylic silanes (ViSP and ViSO) received from Union Carbide
were reported^^ to have both functional groups Si-H and V\
85
s
500
600
Temp. °C
Fig. 11-44. DSC Thermogram of A-PSS-P Showing that the Thermal
Crosslinking Occurs at '-240°C
86
0001
09Z
005 OS" 2
30NV133nd3d %
ooo;?;
CJ
Ln
I
87
Although NMR confirmed the presence of ^ , the presence of Si-H is
not certain. If ViSP and ViSO contain Si-H and ^ functional
groups, they should be crossl inkaole via Pt^"*" even more readily than
allylic PSS because \\ is more reactive than \_ . However, this
was not observed. The vinylic silanes with a Pt"^"*" concentration greater
than 9 x 10"'' mole per 1 g si lane and T > 8U°C for several hours did not
solidify the liquid silanes. This is a question which cannot be answered
unless further work is performed.
The size of the Si-H peak in the IR spectra cannot directly be used
to estimate the degree of crossl inking because of the high bond energy
(-314 KJ/mole).5'+ This is shown by the Si-H IR band at -2080 cm"^ in
SiC after pyrolysis at both 1000°C (Fig. 11-40), and 900°C (Fig. 11-41).
Morterra and Low^z also observed the growth of the Si-H absorption
peak when methoxylated aerosil was heated in a vacuum up to 750°C, while
the absorption peak for the -CH3 stretching band at ~3000 cm"-*- decreased
as the length of heat treatment at 750°C in vacuum increased.
Nevertheless, it is shown in Figs. 11-22, 11-24, 11-45, and 11-46
that the degree of crossl inking appears to be a function of the Si-H
peak size. This is another area that needs to be further investigated.
A difference between vinylic and allylic silanes under crosslinking
conditions is in the reactivity of the functional groups. Vinyl groups
are more reactive than allyl groups by vinyls forming more stable radi-
cal intermediates. This was shown by the lower temperatures needed to
crosslink vinylic silanes. This advantage is somewhat curtailed by a
greater tendency of vinylic silane to be oxidized. Thus, one should
88
|PSS/28DAY
VRAY
Fig. 11-46,
2600 2000 1700 ,1400
WAVENUMBERS(cm')
FT-IR Spectrum of a Region Showing the Fffect of
Crosslinking on Si-H Band Intensities at -2080 cm"
1
89
expect that in synthesizing more reactive silane precursors for cross-
linking there is the danger of introducing none oxygen contaminant in
the polymer and, thus, in the pyrolyzed product.
In Fig. 11-46, the as-received PSS shows a sharp and strong absor-
ption band for Si-H at ~2100 cm"-'-, a medium sized band for y-ray irradi-
ated PSS, and a small band for 10 wt% DCP treated PSS. This means that
PSS crossl inking can occur between Si-H's (Fig. 11-27), as well as Dy
methyl free radicals from OCP at nigher temperatures. The bond energy
of E C-CH^-H is 418.4 KJ/mole63 and the bond energy of a Si-CH2-H should
be a little less than that of = C-CH2-H because of a greater electropos-
itivity of a Si atom than that of C atom. Still, the bond energy of
Si-H (314 KJ/mole)5'+ is much smaller than that of = Si-CHg-H, hence the
crossl inking of PSS by DCP proceeds with Si-H bonds breaking at ~150°C
followed by formation of Si-C-C-Si linkages via methyl radicals at
~250°C. In Fig. 11-27, as-synthesized PSS shows a small exothermic peak
at ~16U°C, which probably corresponds to the crossl inking reaction via
Si-H. The DSC for J-PSS2 in Fig. 11-28 before the DCP treatment shows
negligible crossl inking via Si-H coupling during the heating schedule of
the DSC. Rearrangement of the polymer chain is thought to occur at
~400°C. The small spikes at 100°C correspond to water evaporation. The
reason for sharp endothermic peaks at ~520°C is not known, but it is
thought to be the evaporation of a fraction of low volatility. After
the endothermic rearrangement, decomposition to eliminate H2, CH4, CgH^,
etc. actually begins to occur at ~420°C.
The minimum at ~200°C for PSS with in situ DCP crossl inking (Fig.
11-27) must be due to the decomposition of DCP. Under the heating rate
90
of DSC (5°C/min), the DCP crossl inking reaction may not be able to keep
up with tne heating rate. The exothermic reaction was incomplete until
the temperature was ~220°C. This supports the previous observation of
incomplete crossl inking with DCP at temperatures below ~200°C and the
low TGA char yield of the in situ DCP crossl inked PSS (Fig. 11-37).
In the DCP precrossl inked J-PSS (Fig. 11-23) no chain rearrangement
is evident. Instead of rearrangement, the decomposition begins at a
slightly lower temperature, ~400°C. Tnis may mean that the molecular
rearrangements have occurred during the preceding DC? crossl inking
reaction. Tne primary chain rearrangement is probably a Kumada type, 6"+
as shown in equation II-2.
Me H
I /
= Si-Si = > -> Si-CH2-Si = (11-2)
A pyrolysis GC study of PSS (Figs. 11-25 and 11-26) with DCP showed
a large amount of methane and acetophenone, which are some of the pro-
ducts from the proposed crosslinking reaction given in Fig. 11-47. The
amount of methane is too much to come from the Si-H coupling alone.
Thus, the methane must be formed by the methyl radicals of DCP after ab-
stracting methyl hydrogen from Si-CH2. The possibility of crossl inkage
via Si-Ph-Ph-Si is doubtful because of the greater bond energy for-O^H
(112 Kcal/mol) than for -CH2-H (104 Kcal/mol).55
During crosslinking and pyrolysis, a precursor polymer is decom-
posed and fragmented, preferably with free radical modes to achieve a
high ceramic yield. However, pyrolysis conditions strongly affect the
density of the pyrolyzed product. Density is increased during
91
Ph-C-00-C-ph ^-^A§-^ ph-C-0
Me Me
q
-> Ph-f-ME + Me«
riE
Me Me
f^E CH<
Me» + 4Si-Si47. "^^S?^ 4^1— Si^ + CHi^
Me f^h
>i Ar
■Y J N
Me Ph
Me CH'
Me Ph
2e^i-Si^.. ^ ^i-Si-h.
Me Ph
Me CH-
I[1e Ch2
-fSi-Si-4^
k Ph
Fig. 11-47. The Crossl inking Mechanism of PSS by DCP
92
pyrolysis but not usually to the tneoretical value due to the porosity
generated by evolved yasses during tne process. Figures 11-30 and 11-31
show the uneven surfaces and pores formed on PSS and ViSP after cross-
linking and pyrolysis at 900°C in vacuum or in N2. The blisters formed
on a sintered mass of SiC that are created during the crossl inking pro-
cess are in subnicron sizes and remain after pyrolyzing up to 900°C.
Although the SiC formed from organosi 1 anes and pyrolyzed at ~1000°C
is known to be the beta phase, the x-ray diffraction pattern of the pow-
der is amorphous because of its extremely fine grain size (~3 nm).2'+
Most TGA data for char yield are for samples after being ex situ
precrossl inked in either N2 or in a vacuum. The ^1% char yield of PSS-0
suggests that some of the most volatile fractions escape during the
crosslinking reaction. However, the char yields of crossl inked PSS-0
with respect to the PSS-0 starting weight are in the range of 35-45%,
which is close to the theoretical yield of 45%, according to equation
II-3.'*o
Me Me
-► Ph-H + CH4 + 2H2 + 2SiC (1 1-3)
h Mr'"
This demonstrates the potential usefulness of PSS oligomers as a filler
phase for porous ceramic bodies by infiltrating the pores as small mole-
cules followed by crosslinking and pyrolysis in the pores of the ceramic
body.
A higher char yield of A-PSS than that of PSS without precrossl ink-
ing treatments is thought to be a result of crosslinking between Si-H
93
and S= during the pyrolysis, as discussed previously in this chapter.
This in situ thermal crosslinking also applies to ViSO and ViSP.
In the case of the more reactive ViSP, temperatures above 11U°C
with UCP and above 150°C without DCP are required for crosslinkiny
(Table 11-9). The Petrarch's H-Si-Si-0 substituted POMS was difficult
to crosslink by all methods tried. This raises a doubt that the polymer
has a Si-H functional group, or that a Si-U group may somehow effec-
tively inhibit the functionality. No characterization to determine the
structure was carried out on tne Si-0 linked polysilanes. Other evidence
to increase doubt of the presence of a Si-H group is in the essentially
nil TGA yield of the polymer upon pyrolysis. It may be the Si-0 which
makes the crosslinking very difficult.
The Petrarch's phenyl substituted Si-Si-0 POMS was crossl inked with
difficulty by DCP which increased the TGA yield to 27%. The cross-
linking mechanism of this polymer by DCP is presumed to be similar to
the PSS/DCP system, but a Si-0 group must be suppressing the free radi-
cal crosslinking. The usefulness of this polymer in ceramic applica-
tions is presently unknown, i.e. the effect of Si-0 substitution needs
to be investigated further.
The TGA yield of in situ DCP crosslinked PSS-IP is -30%, which is
lower than the theoretical yield of 45%. This low yield is due to the
constant rapid heating rate (10°C/min) in the TGA operation so that the
crosslinking temperature in the ~200°C region is passed in a few min-
utes, which is not enough time for complete crosslinking.
A pyrolysis study of PSS was carried out by Sinclair's at 400°C in
vacuum. Each fraction he obtained is shown quantitatively below. This
94
distribution of PSS fractions suggests that the crossl inking of PSS via
DCP allows recovery of fractions up to sone of median volatility (54.2%)
as SiC product. The char yield of ~50% is in fact the inaximum theoreti-
cal yield. However, without crossl inking, only the nonvolatile solid
and maybe some of the low volatile solid fraction (~10%) is recovered as
SiC upon pyrolysis.
Fraction
Highest volatility
lost through vacuum line
High volatile 1 iquid
Medium vol atile 1 iquid
Low volatile sol id
Nonvol atile sol id
Wt%
12.4
31.7
17.7 "
28.6 >
7.y J
54.2
As the organosil anes were heated in a quartz tube furnace in N2,
yellow gas was generated and condensed on the colder region of the tube
wall. This same gas corroded the DSC metallic sample chamber, TGA
sample boat made of platinum, Nicalon® continuous SiC fiber in SiC/SiC
composites, and the thermocouple of a furnace. The yellow corrosive gas
was first suspected as a chlorine compound, e.g. CI2, HCl , HOCl , etc.,
formed by the residual chlorine from the starting dichl orosil ane mono-
mers. The residual chlorine was tested by dissolving PSS in benzene and
hydrolyzing any Si-Cl with water by shaking the benzene phase and water
phase in a separatory funnel, then the aqueous phase was titrated with
AgN03 solution. No precipitate was observed. Based on this test, the
possibility of the yellow gas as a chlorine compound was ruled out.
95
Silicon and/or SiC, at low oxygen partial pressure, can form sili-
con monoxide at elevated temperatures. ^ 5»67 jt is then possible that
the yellow gas is SiO, but the chemical reactivity of SiO with respect
to metals, SiC, and SiOg is not known. If it were SiO, the low partial
pressure oxygen source must be in the inert gas, e.g. a commercial re-
search grade No gas contains ~1 ppm O2 and ~1 ppn H^O, and also in the
organosilane itself.
Figure 11-38 shows a large absorption peak for Si02 at ~1073 cm" .
This large absorption band of SiOp is caused by a Si02 film on the sur-
face of a monolithic PSS/SiC sample after crossl inking and pyrolysis at
960°C. This surface oxidation must be caused by oxygen in the inert gas
and/or during handling in air prior to pyrolysis. An FT-IR spectrum of
the same PSS/SiC sample after the monolith was crushed into fine powder
was taken again. This is shown in Fig. 11-39, along with commercial
6-SiC Nicalon®, which is made from polycarbosil ane. The large Si02
absorption band previously shown is diminished to a small hump at ~1U50
cm"-'-, while the bands corresponding to Si-C at -800 cm"-'- remained the
same.
In Figs. 11-40 and 11-42 for J-PSS/SiC and ViSP/SiC pyrolyzed at
900-1000°C, it can be seen that some type of organic residue still
remains, as well as Si-H groups. This may be either from an incomplete
transformation of the organosil anes to 6-SiC or from contaminants from
external sources during handling prior to pyrolysis. This means that
temperatures above 900°C are required to convert an organosilane com-
pletely to B-SiC. However, an FT-IR spectrum of a ViSP/SiC monolith
96
pyrolyzed at 950°C and surface polished with SiC grit followed by wash-
ing with acetone shows an essentially identical spectrun as Nicalon*®
(Fig. 11-43). Therefore the organic residue nust be concentrated on the
surface.
Conclusions
It has been shown that these various organosil anes as polymers and
oligomers can be successfully converted to 8-SiC after crossl inking
treatments and nearly reach the theoretical yield upon pyrolysis. The
process of SiC production from silane monomers and PSS is summarized in
Fig. 11-48. Tne applications of these materials and techniques for
making ceramic composites are the subjects of the following chapters.
The exploratory organosilane (OS) precursors used in this work have
the potential to be formed into desired shapes using conventional low
temperature plastic processing, then pyrolyzed to obtain SiC material
with a yield that is nearly the theoretical limit. These precursors can
also be used to impregnate porous ceramic bodies.
The highest char yield is given by the polymer fraction of vinyl ic
silane (72%). The allyl group on the polysilane chain improved cross-
linkability. A complete crossl inking of OS precursors increased the
char yield of SiC. For complete crossl inking, OCP with temperatures
greater than 250°C for PSS and greater than 130°C for vinyl ic silanes
are required. Other combinations of the functional groups such as Si-H,
— ^ , — v^ , <\ , etc. should further improve crossl inkabil ity and
ceramic yield.
97
>1
^c
1 > i^O I s
■ _j s >•
o
J
o
l_
>>
o
-5
o
1^
c £
CO 1
* I lY
o
a.
o
51^
c
o
'm
o
w
O
X
(M
X
c
u
T3
z
\0
o\ 5;
10 I
1^
•c
o
£
E
o
o
>>
a.
c
o
u
0)
>>
I—
o
S-
>>
a.
-a
en
c
r— Ol
lA T3
t/1 •!-
o ja
C_> ITS
(_>
c c
o
•r- U
-(-> -1-
m I —
N -r-
■r- U~l
01 c
E -r-
>, 03
r— +J
O J3
Q. O
CO
I
o>
98
Char yield of SiC is roughly shown to be a function of concentra-
tion of the crossl inking agent DCP, and a fjnction of temperature and
time. However, temperature has a much greater effect on the degree of
crossl inking than time. The possible oxygen contamination from the
crossl inking agent, DCP was not observed.
These OS precursors are prone to oxidation during handling, cross-
linking and pyrolysis to include oxygen contaminant in the final SiC
product. For a complete freedom from oxygen contamination, there must
be more convenient ways to process the OS precursors, e.g. a glove box
fully equipped to avoid oxygen.
In order to increase the polymer fraction of PSS, a longer reaction
time than two hours would be desired in the synthesis.
Variables that may have significant effects on the PSS polymer
yield and quality are 1) rate of chlorosilane monomer addition, 2)
amount and kind of solvent relative to the reactant, 3) state of disper-
sion and the amount of sodium, 4) reaction temperature, 5) reaction
time, 6) reactant's molar ratio, and 7) environmental factors such as
oxygen, ultraviolet light, etc.
To improve the lot-to-lot reproducibility of the processing of OS
precursors to obtain SiC, more work on the polymerization procedure and
an understanding of all the process variables is necessary to establish
standard procedures.
Although y-ray irradiation can crosslink OS, there needs to be a
source with a higher dose rate than ^^Co to be practical for PSS. A
source such as ^^''cs may hasten the crossl inking of OS.
99
To produce a large dense monolithic SiC body directly from OS,
foaming and gas generation should be controlled during crossl inking and
pyrolysis. In order to control these phenomena, a chemical additive as
an anti foaming agent needs to be found and/or a new polymer with no
foaming properties needs to be synthesized.
CHAPTER III
SILICON CARBIDE/SILICA COMPOSITES FROM CARBOSILANES AND ALKOXYSILANES
Introduction
The significance and advantages of the sol-gel route to produce
glasses and ceramic materials have been given in Chapter I of this dis-
sertation. Although the production of silica glass monoliths via the
sol-gel method is highly significant by itself, the problem of intrinsic
brittleness of glass may be modified by forming a composite.
If the strength of a glass were determined solely by its lattice
cohesive energy, one would predict the glass to be very strong. In all
cases, glasses are weaker than otherwise expected because flaws concen-
trate applied stresses. The problem is flaw sensitivity, causing cata-
strophic failure under applied stresses, and susceptibility to thermal
shock. Because of these problems, glasses have acquired a reputation
for mechanical unreliability.
Many ways to strengthen glasses have been studied, ^8 including the
following:
• annealing
• compressive stresses on the surface via tempering or ion
exchange
• dispersion hardening
fiber reinforcement
• reduction of flaws
100
101
The strengthening nechanisms and effects at room tenperature for
glasses may not be the same as the effects at high temperature. In
metals, conventional precipitation-hardened metal alloys are not ther-
mally stable and the precipitation may coarsen at elevated temperatures
well below the softening temperature of the alloy, resulting in a reduc-
tion of mechanical strength. ^9 In contrast, many ceramics and glasses
maintain strength and exhibit stability at high temperatures witn low
density and chemical inertness.
In an effort to increase strength and toughness, barriers to crack
propagation in the form of discrete particles or fibers have shown some
success, ^^ similar to the fine dispersion of second-phase particles long
used for metallic systems. ^° The strengthening effect in the case of
metals is attributed to various dislocation impedement mechanisms. ''O
However, since dislocations do not exist in glasses, particle dispersion
strengthening of glasses must rely on a different mechanism. Hasselman
and Fulrath^i observed dispersion hardening effects in certain systems
of ceramics by dispersion limiting the size of Griffith flaws, thereby
raising the stress required to initiate or propagate cracks. Lange'72
has proposed a mechanism that strengthening may occur as a result of a
line-tension effect due to particles initially pinning a propagating
crack front and causing a detour similar to that observed for disloca-
tions.
Another aspect of dispersion hardening may be the strengthening
effect of the dispersed particles with a higher elastic modulus than the
matrix. A fine particle dispersion may also inhibit grain growth or
102
crystallization of glass at high temperatures and hence lead to an
apparent increase in strength. Therefore a distribution of dispersed
particles on a molecular scale should yield a large inprovement in the
mechanical properties of a glass.
Using the techniques described in Chapter II, organosil anes may be
introduced directly into the monolithic body of a sol-gel derived silica
glass matrix to obtain a strengthened silica body after appropriate heat
treatments. The organosilane precursors to SiC are thereby dispersed
homogeneously in the matrix of silica gel. Upon pyrolysis, the SiOo
matrix is reinforced by molecularly dispersed SiC particles, thereby
yielding a molecular composite. Developing procedures and understanding
the process are the objectives of this work.
In this chapter, results of the expected hardening of SiOo glass by
Sic molecular dispersion are presented, as well as the fabrication tech-
niques used to produce such a body.
Experimental
Infiltration of silica gel matrix with carbosilanes
Silica gel matrices to be impregnated were prepared by hydrolyziny
tetramethoxysilane (TMOS) at ~90°C with water in TMOS/H2O molar ratio of
1 mole of TMOS per 4-17 moles of water and adding 0.5-2 w/o of acids or
1:10 molar ratio of the acid to TMOS. This liquid mixture called a sol,
was cast in plastic molds and aged at 60-80°C for 24 hours after the
molds were tightly sealed to prevent liquid evaporation. At the end of
24 hours some gel monoliths were allowed to age for an additional 24
hours in the original molds and some gels were transferred to a water
1U3
bath for further aging. At the end of approximately three days from the
beginning, the seals were broken for slow drying in an oven with a
steady increase in temperature from 80°C to 150°C in 12 hours. The
typical size of a dried gel was ~20cc.
The dried gels were decarburized in monolithic forms by heating
them in air to 500-600°C with a heating rate of --65°C/hour and held at
500°C for two hours. At this temperature, partial dehydroxyl ation and
nearly complete decarburation is expected to occur. Tne decarburized
and partially dehydroxyl i zed gels were stored in a desiccator, after
cooling to room temperature with ~80°C/hour rate, until the time for
silane impregnation.
The organosi 1 anes or carbosilanes used to infiltrate the silica
gels were polysi 1 astyrene (J-PSS2), vinylic silane oligomer (ViSO), and
vinylic silane polymer (ViSP). The structures and physical states of
these silanes have been described in Chapter II. The solid silane
J-PSS2 was dissolved in toluene or THF to make up ~20 w/o solution. The
high viscosity liquid ViSP was diluted with THF in 1 g silane/2 ml THF
for impregnation.
The impregnation of silanes into the silica gels was carried out by
placing a monolithic piece of the gel in a silane solution containing 1
vol% of 3-aminopropyl triethoxysilane as a wetting agent and 5 wt% of
DCP and soaking it for 2 to 12 hours. For deeper penetration of silanes
into the gel body, vacuum impregnation for 30 min-6 hours was used. At
the end of infiltration, the infiltration chamber was brought to atmos-
pheric pressure and left for 30 min. before transferring it into a
pyrolysis tube. The processing map is shown in Fig. III-l.
104
>
t-
<
I-
<
_i
o
<
CO
z
o
ti^
(O
2
w
o
z
LlIO
ZUJ
<
cn
UJ
<
m
cr
<
o
o
z
>-
s.
— c:
Ul
<
I-
C/i
O
CO
a:
o
lO
tr
5=>
OliJ
L-J
O
UJ
t-
< ^
u
z
1-
<UJ
ii
-IK
to
O .
zz
§1
lie
_1
o
(T
O
>
X
LJ
3
X
IT3
3
J2
>5
O
Q.
e
o
CM
c
c_)
00
a.
re
I/)
to
01
o
o
s_
3anivy3dW3i
105
The impregnated gels were pyrolyzed in N2 °'" ^" ^'' ^^ various temp-
eratures with the heating rate typically 50°C/nr. The pyrolyzed SiC/
Si02 composite monoliths were stored in a desiccator, after they were
cooled to room temperature from the pyrolysis temperature, until micro-
hardness, porosity, surface area, densities, etc. were measured. Micro-
hardness values were measured by using a Kentron Tester or a Leco Model
DM-100 with 0.5-2 kg load and a 136°C diamond pyramid indenter. Both
diagonals of the indentation were measured and the average values were
used to compute the Diamond Pyramid Number (DPN). Porosity and surface
areas were determined by using Quantachrome' s Autosorb-6. Densities
were measured by mercury, water or propylene glycol volume displacement.
In situ molecular composites of Bulk SiC/$i02 Monoliths .
For a silica gel matrix, TMOS was hydrolyzed using acidified water
in 1:4 molar ratio of TMOS/H^O and organic solvents. The water used
here was previously degassed by inert gas purging to improve the action
of CFRI IJCP. The acids used in ~2 vol% were lactic acid and formic
acid. These are strong organic acids and are expected to catalyze the
hydrolysis of TMOS readily and possibly improve the compatibility of the
TMOS with organosilanes. Solvents used were isopropyl alcohol, THF,
toluene, n-butanol , and amyl alcohol in 100-150 v/o of TMOS. Approxi-
mately 1 v/o of 3-amino propyl triethoxysil ane was added as a wetting
agent to promote the retention of polysilanes in the matrix.
After the hydrolysis was complete (~20 min. stirring by a magnetic
stirrer at room temperature), 0.4-4 w/o of an organosilane (OS) dissolv-
ed in THF (-1/5 volume of TMOS) with 4-5 wt% DCP (wt% with respect to
106
OS) was mixed with the hydrolyzed sol for 20 min. at 30-40°C in a closed
container.
The 0S/Si02 sol mixture was cast in Teflon molds. They were sealed
tightly to prevent any evaporation of solvents and water, followed by
gellation and aging at ~60°C for 12 hours. Curing was followed with the
temperature gradually raised to ~150°C in 6 hours. The gels containing
OS were cured at ~150°C for 6 hours. The temperature was brought down
to 60°C at the end of 6 hours of curing and the seals were broken for
slow drying in an inert atmosphere. Drying was continued for two days
at 60°C in N2 or Ar with a gradual increase of solvent evaporation by
opening the lids. The typical size of a dried gel was ~20cc.
Pyrolysis and densi fication were carried out by placing the dried
composite green bodies in a tube furnace with N2 gas flowing at ~5 ml/
min rate and a heating rate of ~20"C/hour up to 150"C and held at 150°C
for three hours. The temperature was raised to the range of 150-850°C
with a rate of ~100°C/hour and held at 3bO°C for two hours. The cooling
rate was also controlled; 200Vhour to 660°C, held at 650°C for one
hour, 200°C/hour to 300°C, held at 300°C for one hour, then lUU°C/hour
to room temperature.
Microhardness, pore volume, surface area, and density were measured
the same way as for the infiltrated SiC/SiOp composites described in the
previous section. The processing map for this process is also given in
Fig. III-l.
SiC/Si02 molecular composite powder
A powder form of a SiC/Si02 composite was prepared by mixing 0.6-
6.3 g PSS, 8-10 ml TMOS in 10-15 ml benzene or toluene, and 0.05-0.8 g
107
DCP, in an apparatus shown in Fig. II-3 with N2 gas flushing over the
solution for two hours before heating began. After two hours of degas-
ing with N2, the solution was heated to a gentle reflux for 12 hours
with continuous N2 flowing. At the end of 12 hours, the solution color
changed from dull gray to brownish yellow.
The reaction mixture was precipitated in 80/20 dy volume HeOH/HoQ
solvent and washed with the solvent three times before drying in a
vacuum oven at 70°C for five hours. After reprecipitation in MeOH, an
80-90% yield was achieved.
Characterization of products .
Fourier-transform IR spectra were taken by using a Nicolet MX-1
FT-IR Spectrophotometer, SEM micrographs and EDS spectra by a JEOL Model
JSM-35C, NMR by a Varian XL-100 Nuclear Magnetic Resonance Spectrometer,
and by Nicolet's High Resolution FT-NMR, BET surface ared and pore size
distributions by Quantachrome' s Autosorb-6, UV-Vis transmittance by
Perkin-Elmer's 552 UV-Vis Spectrophotometer, IR transmittance by Perkin-
Elmer's 283B IR Spectrophotometer, reflectance of visible light by a
custom built optical microreflectometer. X-ray photoelectron spectra
were taken by using a Kratos model 800. Hot-stage x-ray diffraction
patterns in a helium atmosphere were obtained by using a Philips X-ray
Powder Di ff Tactometer as the sample powder was heated on a platinum
substrate with HTKIO High Temperature Hot Stage by AP Parr Co. of Graz,
Austria. Microhardness and fracture toughness of the monolithic compos-
ites were determined by using a Leco Microhardness Tester and according
to Antis et al.'^^ Flexural strengths were determined by using an
Instron Testing machine.
108
Variations in experimental conditions, reactant amounts, and chemi-
cal additives are summarized in Table III-l for the typical composites.
Results
A photograph of partially dried OA gel monoliths to be impregnated
with silanes is shown in Fig. III-2. The pore size distribution of the
OA gel is given in Fig. III-3. The BET surface areas, mean pore sizes,
and total pore volumes of typical composites are summarized in Table
III-2. The volume loading of SiC in the pyrolyzed composites are esti-
mated from the TGA char yields given in Chapter II for each US and the
volume OS added to the sol. Approximately 3 vol% SiC was maintained.
Repeated attempts to produce edge-notched SiOo matrix monolithic
specimens for fracture toughness were unsuccessful. Attempts to heat
the composites higher than ~850°C in nitrogen also failed due to the
foaming of the matrix.
A photograph of an Si02 monolith strengthened by in situ bulk de-
composition of an organosilane along with a pure gel matrix and gel with
OS before thermal treatment is shown in Fig. III-4.
Infrared transmittance curves of SiC/Si02 glasses with a fused
silica glass as a control are shown in Fig. Ill-b. A reflectivity curve
of a bulk SiC molecular composite glass is compared with a blank gel
glass in Fig. III-6. Ultraviolet-Vis transmittance curves of the compos-
ites are compared with a blank FA gel glass in Fig. III-7. Figure III-8
shows IR reflectance curves of FA gel and ViSP/FA gel. An enlargement
of the band at ~800 cm-1 is shown in Fig. III-9.
Fourier-transform IR spectra of PSS-10/104 composite powder before
and after pyrolyzing at 900°C and at 1270°C for three hours are shown in
109
u
(/I Q
O •
CL— '
E
O
o
Q.
00
I I
o o
O Q_ CO
OO OO CO
•r- -r- Q.
> >• -D
I I I
=c =1 <:
CO
0)
-»->
•^-
TJ (/I
C O
ra Cl
e
" o
en CJ
■!->
C CM
3 O
o •.-
E oo
cC -^^
o
■M -1-
C t>0
fO
4-> i*-
O O
na
Ol CD
Qi E
o
• CO
•
(/)
t — t
c a;
1
O M
*—*
•1 — -r-
•— 1
-(-> (/)
*—*
•^ OJ
-o -C
<D
C +J
r^
o c
J2
o >,
"3
CO
1—
r—
fO o
-(-> ■!->
c
<D -O
E QJ
■1- (/I
t- rD
a;
Q. (/I
X a;
LU >
•^
>4- +J
o ■<-
■a
>i-o
t- <
TJ
C , —
E "3
3 U
00 -r-
E
<U
-C
o
o
4-) Ol
i- ITS
•— o
o c
T3 O
O E
-c <:
■I-)
a;
0)
o.
(U
t-
3
(/)
CO
Ol
ns
OJ
CD
C
E ^
O (O
o o
t_ CO
o
00
a.
CO
3
O
>
c
o
E 13
<D C
OJ
E t-
O Q.
O E
t_ T-
^ CO
O Q- 00
00 CO Q-
•I- -r- I
> > -3
(U
PI
CO
O
CL
E
O
OJ
C3
o
T3
OJ
O)
u
o
-a
o
ro
Q.
oo
O
X
<:
oo
o
OvJ
o
o
s:
31
Q.
(U
2:
CVJ
1—
•
O
1—
3:
<V
^—
Q
'o
OJ
o
£
<u
(U
r^
'o
E
O
r—
o
E
1 — 1
'0
E
Lf)
i-H
•
2
o
•
E
E
1— 1
r-H
o
in
o
t-H
CTi
n
10
OO
■^
LU
^
■t->
1—
E
u
Q.
t_
n3
■=t
r^
4—
O-
(U
r—
—
*©
E
E
E
LO
^■^^
.—I
r— I
2
•
•
•
>*
no
3: 3: o
O O r:
3 = ca
aa CO ■~>,
O Q. 00
00 OO Q.
> > ^
^
:e
c
c
c
0)
OJ
0)
Q.
Q-
n.
t/1
Q-
CO
IT)
oo
Q-
• r—
•^
1
>
>
'n
r^
^
^3-
CM
■^
CM
CM
a^
(Ti
1 — 1
r-l
1
1—1
^H
VI
t — t
U^
1
00
1
oo
00
O-
00
Q.
OO
1
ty)
1
Q.
<
a.
'O
00
o
00 00
o o
00
o
00 oo
o o
CM CM
Lf> ^ ~~>,
CM 01 ai
cnLD i-o
^00
S E
E
■-0
tn
CM
CM
uo
^■^^
'■'-V,
CM
cn
en
01
LO
to
•
•
1— 1
■4->
C
o
o
cu
00
^—
Q. 00
xi
C/O 00 Q.
la
•r- -1- 1
h-
> =>
(/I
o
Q.
E
O
C_3
CSJ
o
00
00
3
+J
•^-
00
CM
00
</>
Q.
3:
1
"^
L.
O.
s
Q.
^
N
00 00
C
•r— -T—
Ln
OJ
>■ =>
CM
-O
■0
XI
,_
-0
•1—
■0
•^
-C
(O
fO
<t3
(O
•f"
•^-
n3
•^
4->
E
•^-
E
I_>
:e
:_
-(->
,—
^
fO
lj_
>,Ll_
00
r^
3
3C
oo
4-
(O
E 3:
>+-
CQ
1—
rtj 1—
s:
CM
a;
Q.
s:
CMOJ
Q-
1—
JZ 0)
f^
OJ
OJ
1—
T~ ^_
OJ
<D 01
f—
^—
r—
Q
Q
a;
a>
E
OJ
<D E
'0
'0 '0
r—
r- E
B
E
r—
E
E E
LO
*"*^
Ln
E
E r-H
LO
f-H
s
E
E
. — 1
«;)• .—1
3
.— I ^ O O O O VC
1— I «d- O O O O i£>
C3
•
Ol
C
U1
^-^
<D
e
0)
Nl
OJ
c
c
a.
-0
<D
a;
O.
3
N
^
<->
a
C
Q
c_
OJ
^
O)
^
E
cn
a;
LO
4->
,—
^
r^
• r—
c
1—1
ro
•
to
^-^
•
i-H
Q.
1 — 1
1
I-H
E
■^^
00
r)
00
00
.-H
00
u
1
Q.
1
1
'y>
1
t-
00
•a:
tn
'D
(O
00
Q.
^—
a.
zri
Ol
3
CD
U
CD
oo
,_l
OJ
•
CM
•
^—
ro
1^
. — t
LO
"■"^
'■'-^
- —
.— 1
y
Ln
00
00
^^
00
J s:
s:
2:
h-
1—
1—
s
•r-
h—
oo
^—
^—
r—
•v*^
E
e
E
^^
E
•^-
00
1 — r
r— I
■—1
CO
Ill
Fig.III-2. Silica Gel Matrices After Aging and Drying
Prior to Decarburation and Impregnation.
112
0.03
1000
PORE RADIUS, A
Fig. III-3. Pore Size Distribution OA Gel Showing no Change in
Average Pore Sizes After IBO'C and 500° C Heat Treatments
113
<4-
O
E
(/)
3
3
>
'o
>
o
<u
a.
o
a.
OJ
■tJ
1 —
f=
■!->
o
o
<*-
»>
T3
<U
'^*
■(->
•r-
T3
(O
n3
3
O
o
O
Q-
>1
c:
4J
1/1
OJ
•%
Q
a;
0)
3
{_
I—
<D
»a
o
><-
3
oo
to
1—
4^
LU
CQ
1/1
o
Q.
CM
E
O
1
o
t— «
C\J
o
13
O
r-
>1
-!->
■t->
m
■ f—
i_
CO
o
c:
<D
OJ
f—
3*
Cl
•=^ O lD IX)
r^ ro "^ r~-
r-^ =S- 3
oi -^ CM
I I
30 O
CM OO
o
a.
CM CM lo
r-< ro CM
CM ys ^
CM
1— 1 LD '^
ZO CM
n
O -O JT)
'-n ro
QJ <<
o
««
ID.
1/1
CM r~-
3
• • • •
c
• r—
00 'S- CO fVJ
"3
"O
.— 1 1— 1 r-l r-l
a;
fO
2:
q:
CO
1^ o .x>
t— I .r> <£)
Lf) 1— 1 in
00
CO
00 LD CO ^o
00
■^
.— 1 ^ ro >— 1
. — 1
<s-
-O CO ^ .— I 1^ CM LO
■^ 1^ • i-o ^ •— I 00
ro CM 00 CM .—I r-~ .— I
c
OJ CJ>
E o
■(->
113 •
OJ h-
<u
0000
O O iXi o
CO i»0 r^ 00
o o
CO
f—^
000
Ln
'-n
un
CO
00
00
00 o^ 00
00 CO
CJ
CO
o
Q.
E
O
o
csi
c
C_)
c
"T"
3:
lE
:r 3r
c
c
000
Q-
O-
00
3
3 <u 3
u
1 — OO 00 oo
OO
00
00
ca
CO Q. 20
a.
OJ .,_ .^ .,-
• 1—
■ r-
2.
^-^
— -* ^"^ ^"^
^^.^
C7)> S> >
1 1 1
>
1
>
1
1
00
n. a. Q.
00 00 C/1
00
to
<: <: «a: <
<:
<:
=c
■ r-
■r— -r- 't—
a.
0000
Ll_
U-
>■
>■ => =>
'^
114
J-pss/SiOjgel Composite
White 120°C
Blacl( 800°C
Clear Matrix Pure Gel
t
i|i|Mi|l|l|MM|i|i|!mr|Mi|MMMi|MMMMMqMi|MifMiJMi|'ti|'|il'IMi|i|'lM'|i|'l'|ili|ilMl|i|'|i
'lt!|ll ^l l|llll|if ll| llll^l ll ( ll (l l l ill )ll l ^llll|lfl l |ll ll | l l iy ( l |^ ^ ^ ^
' — OK GBS OB OU 001 081 081 Oil 08t 091 On QSl 081 OIL 001
|i|i|i|i|i|i|r|i|
Oqc 068 088 QZ2 08Z
llllllllllllllllllllll
llllilllllllllllllllllllllllllllllllllllllllllllllllllilllllllllllllllllllllllllllllllllll^^^
Fig.III-4. SiC/Si02 Molecular Composite from PSS/SiOo Gel
Before and After Pyrolysis With a Pure Gel
Matrix for a Comparison.
115
o
O
o
in
00
o
'
.-1
in
o
<
r-(
M
u
ra
-
u
^^
to
o
r-H
•r^
11
D
un
-^
^i:
00
'Jl
<
b
^
o
[^
•'-'
^-1
rH
-o
a.
3-
0^
ji
C/l
S
-/
o
o
o
. o
; \
i
I ■
1/-
o
o
o
(U
o
T3
c
to
(U
o
00
o
- c
—
TD
c
1
<U
E
to
o
l^ >
1^
o
ft3 C
s- ro
•4-> (
o
u ••—
c
oi on
ittance Sp
ed With a
o
E -t-"
00 n3
o
C C
o
R Tra
mpreg
1 %
O)
115
CD
■a
<u
+->
(O
c
C7)
O)
S-
Q.
E
a.
n3
0)
n3
u-
o
o
o
CO
VI
<u
■M
•r-
<T3
■l-l
•r-
-o
>
O)
•r—
M
+->
>,
U
r—
tu
o
r—
u
M-
>>
d;
D-
IC
O vC ^
117
10
ax
— T ; —
\
1 "^
J
\
\
0)
5C
1
\
LO
<
\
\
cc
U.
'
U
<v
\
\^
o
r-^
ao
' "^
^\>.^^
o
aj
01
[
"■ ^V ^N^,^^
o
o
u
"^ - ^;r^''"^
:». 00
Q.
\
x:-^^
<
-
1
'\
^
0^
a.c
"\
'^
c/: o
^\
<
■H
•rt u-l
/
\
a.
>
> 00
V
•V
1
\
-^--' — ^
1
\
•<
1
/
^
-_
1
;
' j
1
\
\l
\
;)
■
\
/ f
\
\ 1
_>
'^
— ^
\
* ^^^^
--
<
--■~-^
... ."r-t-ras
■
1
"*C
^^
\ 1 \
^ 1 ' L__
c
c
c
OC
C:
O
X
1 %
CO
o
o.
E-
o
CJ
to
(U
-a
c
re
CJ
CO
■•->
<J
<u
a.
CD
I
I
iZ
118
o
CM
UI
0)
C
1/1
o
>1
L/^
^—
o
S-
>i
c
: '
i-
CO
(D
-!->
tl-
<
O
c
r—
t-H
OJ
—
o
CO
T3
C
n3
C3
E <
1T3
o
(U
D.
00
(U
O
c
+->
(J
OJ
I
CO
I
aDUBaoaxjBH %
119
ViSP/TMOS
TMOS gel
800 ,
v; cm '
500
Fig.III-9. Enlarged Spectra of Fig.III-8 Showing Greater
Peak Area at --800 cm'' For the SiC/SiO, Composit
2 Composite
120
Figs. III-IO and III-ll. An expanded IR spectrum of J-PSSl/1120 with as
received J-PSSl for comparison is shown in Fig. III-12.
Proton NMR spectra of A-PSS-0/927, PSS-10/104, and J-PSSl/1120 are
given in Figs. III-13, III-14, and III-15. The TGA char yields of the
powder composites are shown in Fig. III-16. X-ray photoelectron spectra
of PSS-10/924, A-PSS-0/927, and ViSP/BuOH composites for silicon after
they have been pyrolyzed are shown in Figs. III-17, III-18, and III-19.
Scanning electron micrographs of fractured surfaces of ViSO/BuOH,
ViSP/BuOH, and J-PSS/BuOH are given in Figs. III-20 and III-21.
The results of a diamond point micronardness test for SiOo gel
impregnated with silanes are given in Table III-3 and Figs. III-22 and
III-23 for DPN vs. pyrolysis temperature. An optical micrograph showing
a diamond pyramid indentation on an impregnated composite is given in
Fig. III-24. Results for the same test for bulk in situ SiC/SiOo mole-
cular composites are given in Table III-4 and in Figs. III-25, III-26,
and III-27.
The results of fracture toughness measurements directly from an in-
dentation crack are given in Table III-5. Mean value of 5-10 measure-
ments were used in the calculation of Kjp. The formula used to convert
the crack lengths to Kjq described by Antis et al.'^^ -j g given below in
Eq. III-l.
Kj(, = 0.016 (^)^^^ ^-372' ^^^"-^
where: E = Elastic modulus in pascals
(c/2)3/2
P
H = Microhardness in pascals = 5-
2(a/2)''
P = Indentation load in newtons
121
CO
O
s-
D-
s_
C
fO
OJ
s-
o
M-
CO
o
r
4->
O
(U
c
tn
cc
t
o
I
265 £9> S2£
30NV103U3d %
90-2 S
122
t.
u
CVJ
•I—
CO
o
s_
>»
i.
o
1^- °
1/1
Ol.
E
3
+->
O)
Q.
■3ali»a %
123
m
l8-b2
■-ZZii-
o
o
o
m
- E
t/0
u
a.
•^Jt^
1
•-D
(/)
cr
"O
oy
fO
in CO
o
CVJ2
(NJ
— ID
2
^
UJ
c
>
oo
<
£X
O^
1
o
<+-
lO
o
ra
&-
4->
U
cu
Q.
I/O
o
m
)— 4
h-
1
0082 6112 SCt'l
33NV103nd3d %
ZSZ
CJ
124
QJ
CD
o
s.
I
o
cn
3
O
o
I
CO
oo
t
o
E
+->
u
o
+->
o
i-
I
125
CD
1/5
</)
o
s-
1
o
I
o
c
en
c
s
o
o
CM
CO
D-
1
+->
U
(U
c
c
o
+->
o
i_
c
O)
125
a.
a.
CD
CO
(/)
o
s-
III
•1 —
Ul
I
o
I
_CVJ
C3
I
CD
O
o
to
to
o
E
3
S_
-1-1
<u
C/)
c
o
■•-»
o
127
<T
rsi
<•
CN
ON
o
a>
C
r-H
o
1
O
1
1
m
0^
Ul
V.
1
C/1
<
1
' 1
1 1
1 1
1 I
; ;' /
1
■ 1 1
1 1 J
/'' /
r ' /
./ y
-''"^^^ .^^^
^^''"^^ ^^^^^'^^
/
t
^ —
f ■ •
' -^J^
""^
■'^f
\J}
f
s-
Ol
3
o
C-
o
c
o
S-
(J
o o
o -"^
O • cvj
X -r—
•f—
o
o
o
-a
CJ3
I
o
o
o
o
00
o
o
o
(%)3ripTS3y' • 3n
128
SINOOD
129
E
o
<:
en
c
•I —
S
o
CM
CTi
--- <U
C t:
I -r-
Ul ^
U~l i.
Cl. rt3
I C_^
-C
QJ C
(/) QJ
O X3
CL'i-
o e>
</)
CSJCC
•r- t/1
CO c
■^ o
C_) (J
e o
s_
c
■M
o
(->
QJ
+->
Q.
S-
OO
o
Q.
OO
O
D-
s_
X
Q-
00
1
sxNnoo
130
600
400
200-
120
Fig.III-19. XPS Spectra of SiC/SiO^ Composite ViSP/BuOH
Proportions of Silicon and Oxygen As Oxide
and Carbide (top). Oxide and Hydroxide on
the Surface, Respectively
131
Fig.III-20. SEM Fratographs of Bulk in situ Molecular Composite,
Top: ViS0/Si02, Bottom: J-PSS/Si02
132
MJJJf.'^f.Jif"
v'^^^f
'■.■ •- ?,v i>
^v 25KU "X20^0.0* i' :t2B2S%^^^
'•f^?:*^'i8!^^-«^i5:
- ■ . . - . ,;*■ \* ■■-■■.;-. ..
■ , A ;-' ■-■ . H-;
;=^V?*f'^■■>^>> I ••-■'■. r^^'^:V.'^ -
. c - ■ ' -•
.. ;;.'vi;;. ■; :
»■ " ■■ -. •■•■;- - ■'. ■!■'.■?■-.'■ '"".;■<■.■■>;•■'.'. ■
' "' .- '
".■■'-.■■■'*;-■
^^isKy' X200B0
050i*pj.:-l/0U UF^^^
Fig.III-21
Fractured Surfaces of ViSP/Si02 Gel, a Bulk
SiC/Si02 Molecular Composite Cefore(top) and
After Pyrolysis at 830°C
133
Table III-3. Microhardness of Organosilane Infiltrated
SiC/Si02 Composites as a Function of Pyrolysis Temperatures
Composites
and Control
DPT^
b20°C
600°C
700°C
800°C
900°C
OA-ViSO
139±1
233±15
239±14
669±173
814±200
OA-gel
146±58
157±1
174±25
210±51
3U5±103
FA-ViSP
192±55
342±75
35U87
403±46
550±180
FA-gel
129±64
171±24
210±51
291±67
134
1200
1000
800
n OA Gel Control
OGei/viSO
600
400
20C
i
^$'
{f
'/:-^---^
,0
()
[]
n-
500
600 700 800
Temp, **C
900
Fig.ni-22. DPN As a Fuction of Pyrolysis Temperature for
OA Gel and OA Gel Impregnated l.'ith YiSO
135
7oa
60C-
50
400
300
20
00
/
^
AFA Gel control
OGel/viSP
()/
/
/
()
/
/
/
<^ /
o
A
/
/
y
J^
y
y
500 600 700 800 900
Temp^ ° C
Fig.III-23. DPN As a Fuction of Pyrolysis Temperature for
FA Gel and ViSP Impregnated FA Gel Composite
136
Fig.III-24. Indentation Cracks Formed by Loading of a
Diamond Indenter at 120C g on FA Cel/ViSP Im-
pregnated and Pyrolyzed at 8Q0"C, Mag.:200X
Photo Taken After 3 weeks of Indentation
137
Table III-4. Microhardness and Density of ViSP/BuOH, a Bulk
SiC/Si02 Composite as a Function of Pyrolysis Temperature.
Temperature (°C) p, g/cc DPN
600
0.8±0.1
250±50
700
0.95±0.1
350.8±134
800
1.20+0.1
500.0±100
350
1.45±0.1
551±169
900
1.75±0.1
366.3±250
138
0.5
600 700 800
Temp/c
900
Fig.III-25. Density as a Function of Pyrolysis Temperature ■for
Bulk ViSP SiC/Si02 Composite as Compared With Pure
Silica Gel Matrix
60C-
50C-
139
/
/
/
/
Q_
Q
3oa.
/
/
/
/()
/
/
/
<3
U'
20
i
00 700 800
Temp , °c
900
Fig.III-26. DPN as a Function of Pyrolysis Temoerature
for ViSP/BuOH Composite
140
600
500
0.400
Q
300 -
200
OviSP/siOj
K3^
^>-
80
1.0
-^
1.6
12 1.4
Density, g/m I
Fig.III-27. Relationship Between DPN and Density of ViSP
Bulk SiC/SiO^ Composite and SiO„ Gel Matrix
.8
141
(O
«s
■D
e
<-}
^
o
c
i/i
k_
VI
TS
5
o
o
•(—
OJ
Q
O
O
.-o
Ll_
t_
Ll_
T3
OJ
«>
c
CO
•^~
CO
o
O
1/1
-l-J
S_
''~
o
en
O
X
E
O
•
O
LD
00
1
o
»— 1
OO
OJ
o
X!
oo
<a
Q.
O
o o
00
J1
3 O O O --I
00 CM
o ^
ITS
OJ
po OJ ro
"=f
>*
oo
Q.
OJ •
• • • LO ^
•
•
• «s-
s:
. o
o o o • •
o
o
o •
O -H
+1 +1 4H O O
+1
+1
+1 o
^
+( o
r^ CTi OJ -H HH
l^~
OJ
LD +1
o
^ a^
<o ^ '-n ^ r-<
CTi
^
"=3- n
r-l O
O O O r^
fO
(T) OJ
OJ ^
a.
+1 +1
f-H
1 — 1
OJ +4 +1
O^J
c:
CO ^
+1
+1
-H 3D LD
+1
• •
00
T-H
'^ • •
ro
•*
OJ o
•
•
. o o
•
OJ ^
r^
LD
t^ OJ r-H
OJ o^ r--
o^
E
+1 +H
r-H
O
r-H -H -H
3.
^ OJ
1— 1
(-H
4H ro oo
00
• •
+ 1
-H
T— t • •
+1
M
Tl" O^
o
x>
O -^ O
o
lO
a^ '.O
33
Ol
T— 1 r~. iD
1 —
•
O
+1
1^
+1
ro -^
+1 +1
LO
+1
+1 •
r^ 00
LO ^
+t +1
( — I . — 1
-D r^ ^ m ^
•n
«*
•^ LO
• •
+1 -H +1 -H -H
+1
+1
+1 +1
^H .—1
-H »3- 3^ (Tl X)
1-D
ro
r^ ro
"^ ro
-O ^ ro OJ ^J
^
^
ro ro
T3
O
a.
CD
OJ
>>
o
O
3
3
<*-
o
O)
3
I_
0}
■t->
-a
c
ro
(U
o
c
<-
•-n
o
lO LO iT) LO
O O O r-( o
lO
«3
.— I ro OJ o
OJ OJ 'S- CO
ro
o
LO
o
o
o
CO ^ OJ >— 1
'T
"=3-
o OJ
^H OJ P~ Oi
O^
i —
00 CO
J3 ro
O C_3
0) o
O
w) -t-
O O
c —
O O
OJ -r-
m CO
T3 CO
0)
r~- r^
^
>,-c
dl C_5 t_5 OJ
Q.
.— 0)
010 O)
E
1 — on
o o
ro
3 3
<: o OJ cc
to
Ll_ >*-
o LO CO u-
t_3
CJ
OJ
•
o
3
o
•)-> O
-c O
4->
• r—
r— OJ
OJ OJ
•^■
to
•— 00
.^ CO
cn
^^
4-
ro
o
C "
o -
c
•^-
• r- r—
CO , —
• r-
to
OJ
OJ
CJ
O CD
Q. CT1
o
^ o
CO
to
U-)
,— o
■.- cC
••- «a:
• r—
3 OJ
5» U_
> o
>
j3 CC
^J
cn
Q-
ro
•
■o
o
0)
CO
+
3
a. 0)
««
c^ 3
>-
• f^—
Z
^H ro
1 >
«
.— 1
CT
-^ c
C
ro
-^
o<D
c
ud E
c
o
II CO
C_5
0)
LiJ -(->
*
o
to
31 C
-^
C (U
i-
•.- T3
O
CO
2
3 l-
OJ
to
>>-M
to
-Q 4->
ro
0)
r—
T3 r-
ts
OJ
-l-J ro
C7>
ro
c
E <V
• r—
•1- >
C
-U O
I_
on .Q
o
UJ ro
o
142
a = Diamond point indentation diagonal length
c = Extended crack length
The results of flexural strengths of these composites obtained by
using an Instron machine on 3-point bending are given in Table III-6.
The detailed testing method will be given in the next chapter.
Hot-stage x-ray powder diffraction patterns for A-PSS-0/927 powder
on a Pt substrate are shown in Fig. III-28.
Discussion
It has been demonstrated that incorporating an SiC phase into a
pure Si02 glass matrix increases the strength of the composite material.
However, the art of making large monolithic composites of SiC/SiOo via
the sol -gel technique is in the successful fabrication of the Si02 glass
matrix. In this work, it was found that the addition of organosil anes in
a TMOS sol makes it more difficult to obtain a large monolithic glass.
Despite the strengthening effect by the SiC phase, the increasing diffi-
culty of monol ithicity with increasing loading of SiC in the composite
appears to be a combination of three factors:
1) the evolved gaseous products from pyrolysis of the organo-
silane creates flaws and pores (Fig. 11-30),
2) the SiC phase strongly hinders viscous flow of the glass
matrix, thus the necking is curtailed (Fig. III-25), and
143
Table III-6. Flexural Strengths of
Organic Derived SiC/SiOo Composites.
Sampl e
P, g/cc
P, N
Qfiex' MP^ ^flex/p
12 hr soaked
OA gel in ViSO
820°C
1.26 ± 0.1
31 ± 5
4.4 ± 1.5 (3)
3.5
Dried OA gel
150°C
1.17 ± 0.1
67 ± 6
71 ± 6
62 ± 4
25 ± 3 (3)
17 ± 3
15 ± 3
21
HCl gel
2 hr ViSO soaked
820°C
1.75 ± 0.1
533 ± 12
73+6 (2)
42
HCl gel
820°C
1.76 ± 0.1
408 ± 17
60 ± 8 (2)
34
FA gel soaked
in ViSO 10 hrs
800°C
1.34 ± 0.1
62 ± 9
9.2 ± 2 (3)
7
OA gel soaked
in ViSP 6 hrs
1.59 ± 0.1
53 ± 8
25 ± 6 (3)
16
OA gel soaked
in ViSO/ ViSP
7 hrs
1.62 ± 0.1
67 ± 9
6 ± 2 (2)
6
ViSP bulk
Molecul ar composite
0.91 + 0.1
53 ± 10
5.8 ± 2 (3)
6.4
* The numbers in the parentheses denote the number of specimens tested.
144
n 900°
c
npt
/
SiOz
Pt
^
I ..j^J\J
\
WV^vr^VVw/V "T;Vr ^-;/V^^^^^.v^^
20
26
Si02
^~*vv/*.
32
1
38
Pt
44 50
1
80
Fig.III-28. Hot-Stage XRD Powder Patterns of A-PSS-0/927 Composite
145
3) a mismatch of CTE between Si02 (0.5 x iO'*^ in/in °C) and
SiC (4.7 X IJ"^ in/in °C)'^5 increases residual stress (Fig.
III-24).
Although these may all be related, the first factor has been discussed
in Chapter 2 and it seems an inevitable phenomenon for this type of
organosilane. The second factor was evidenced by densi fication behavior
as a function of temperature (Fig. III-25). Gel derived silica without
a SiC phase densified at a faster rate and at lower temperatures. The
main reason for the sluggish densi fication behavior must be in the
action of the SiC phase hindering the viscous flow of the glass matrix.
However, the effect of atmosphere in densi fication behavior may have
some effect on silica viscous flow, primarily due to water content in
the atmosphere.
The third factor that was shown to be harmful to obtaining mono-
liths could give a positive effect on the strengthening and thermal
properties of composites by introducing a compressive stress on the
surface. ''5
An additional factor which has been repeatedly observed and men-
tioned previously is chemical reactions between the SiC precursor OS and
the Si02 precursor gel during the pyrolysis process in a reducing atmos-
phere. This is an important phenomenon for fabrication of successful
composites. Gels heavily impregnated with OS always came out like
crackers: "cracker effect," after the OS/gel composite has been pyro-
lyzed in nitrogen atmosphere. Although the bodies, many times, general-
ly remained monolithic, a severe spalling and/or pores were generated.
146
As a result, the flexural strengths and densities were significantly de-
creased.
The origin of the "cracker effect" may be thought to be from two
sources. One is the anticipated outgasing by the decomposition of OS
and elimination of any impurities during the pyrolysis. Tne other
source is the actual chemical reactions between the reinforcing phase or
the precursor of the reinforcing phase and the matrix phase as shown by
Eq. III-2 for an example in the presence of impurities such as residual
carbon, silicon, oxygen, etc.
2Si0o(S) + SiC(S) > 3SiO(g) + CO(g) AG = -70Kcal /mol ^7 (III-2)
^ A
In this reaction, it can be seen that the glass network is destroyed as
a result of the reaction. This reaction is farther discussed in the
next chapter with respect to oxidation of silicon carbide.
However, as shown in Figs. III-22, III-23, and III-27 and Tables
III-3, III-4, and III-6, lightly (~2 hours soaking) impregnated compos-
ites exhibited a significant increase in microhardness , c7^]gx» ^"'^ *^IC
with little increase in density.
The determination of Kjq by direct measurements of tne crack length
has not been universally established. The method is valid only for the
test specimens which behave normally, i.e. give a well defined radial/
median crack system. ^3 Some brittle materials, e.g. glass, have shown
to give lower Ktq values than values obtained by more standard methods,
such as single-edged notch beam test.^s These materials have shown slow
crack growth well after the diamond point indenter is removed from the
surface of the specimen. Moreover, the method^s has not yet been proven
147
as a valid method for composite materials. This is reflected in the
relatively large uncertainty in Kj^ values (Table III-4).
The KjQ to p ratio of bulk ViSO reinforced Si02 gel (-1.3) is near-
ly as large as or larger than hot pressed 20 v/o SiC/borosil icate glass
composites by Gac et al ."^^ (Kj^/p = 1.6), by Samanta and Musikant'^^
(Kjq/p = 1.2), and much larger than sintered chopped Si02 fiber/Si02
glass composites of Meyer et al .^o (Kjq/p = 0.48). The ratio of ~1.4
for ViSO/ SiOo 9el composites in this work is also superior to hot
pressed 42 v/o SiC/Al203 composites by Cutler et al.8i (Kj^/p = 0.80)
and comparable with zirconia toughened alumina composite by Lange et
al.82 (Kj^/p ~ 1.4).
The flexural strength of lightly ViSO impregnated HCl catalyzed
Si02 gel is increased (Table III-6) to 73 MPa from 60 MPa for the blank
matrix. This is more than a 20% increase in strength. However, because
of the statistical nature of strength, the significance of ~20X increase
is in question and it needs further work. Optimization of the soaking
time in relation to pore size and distribution is suggested for the max-
imum increase in af]px i" future studies.
In order to improve the monol ithicity of the composites, procedures
in gel matrix preparation may need to be changed. Zarzycki et al.83
suggested that larger pore sizes promote monol ithicity of gel glasses.
Yu et al .8'+ supported the suggestion by experimental results. They
showed that acid catalyzed gels always produce smaller pore sizes than
base catalyzed gels. They successfully densified base catalyzed gels,
but not the acid catalyzed ones. This suggests an advantage of the col-
loidal route to obtain a monolithic glass.
148
The acid catalyzed gels used in this work produced cracks and/or
foaming at temperatures above ~850°C. As shown in Fig. III-3, the ;nean
pore radius is in the range of 15 A. This is very snail for gases and
hydroxy! groups to leave the gel structure before pore closure at the
surface. In addition, larger pore sizes should increase SiC loading by
OS infiltration. However, it is shown that larger loading of the SiC
phase is not only unnecessary but also detrimental to the mechanical
properties of the composite.
If a successful consolidation of the Si O2 glass matrix monolithic
composite can be obtained all the way up to full density, the hardness
and strength of the composite body is expected to be much higher.
A J-PSS/Si02 bulk molecular composite, as shown in Fig. III-4 after
pyrolyzing at 800°C for two hours, still maintained a low density of
1.32±0.1 g/cc. Heating above 800°C usually caused foaming. At this
stage, to prevent foaming, use of a vacuum or a reactive gas atmosphere
such as H2 may be helpful. Composites made this way contain no more
than 4 wt% SiC phase in the gel matrix.
An SiOo gel film heavily impregnated with a SiC precursor shows
strong IR absorption in the range from 2.5 ym to 50 ym wavelength region
(Fig. III-5). In the region of 3.8-4.6 urn, a bulk composite with a low
level of SiC (~2 w/o) is more IR transmitting than a fully densified
fused silica control. Above 5 um, they both are IR absorbing.
The reflectivities of SiC/SiOg bulk composites from J-PSS2/TM0S in
the visible range (Fig. III-6) are approximately the same. The curves
above 820 nm are extrapolated by a computer and they may not necessarily
149
be accurate. The clear Si02 matrix shows -2% higher reflectivity than
the black SiC/Si02 composite. The low reflectivity of the gel matrix is
caused by high transmi ttance and that of the black composite must be
caused by strong absorption by the black color.
In the UV-Vis light range, the transmission characteristics of
these chemically derived materials all absorb UV strongly and transmit
visible light. The curves for the impregnated composite and the bulk
SiC dispersed composite are essentially identical (Fig. III-7). The
spikes at 720, 400, and 340 nm for the composites must be due to the SiC
phase.
The infrared reflectance spectrum of a molecular composite ViSP/FA
gel, as shown in Fig. III-8, is almost identical with pure FA gel,
except for small absorptions at -2500 cm"^ and -3700 cm"^ for ViSP/FA
gel caused by adsorbed CO or 002^^ and by adsorbed water, respectively
(see also Fig. 11-42). The absorption peak at -800 cm"^ is mainly
attributed to an Si02 tetrahedral response. However, the area under the
peak for an SiC/Si02 composite is -20% greater than a pure Si02 matrix,
as shown in Fig. III-9. The unique absorption band for SiC is also in
the region of -800 cm"^. Hence, the greater peak area for the composite
must be due to the SiC phase in the composite.
Figures III-IO and III-ll show a somewhat larger proportion of the
Si02 phase than the SiC phase in PSS-10/104. This agrees with the in-
tended proportion, assuming a -30% SiC yield from PSS-10. The propor-
tion of SiC and Si02 phases are in the range of 40% SiC and 60% Si02 in
the composite fired at 1270°C.
150
Figure III-12 shows the structural changes from J-PSSl to J-PSSl/
1120 by the reaction PSS, TMOS, and DCP. It is similar to the cross-
linked PSS in Figs. 11-20 and 11-21. Proton NMR reveals that A-PSS-0/
927 (Fig. III-13) is crosslinked via =Si-0-CH2 by the action of Si(0R)4.
However, the NMR spectrum of J-PSSl/1120 does not show a C-O-Sin linkage
(Fig. III-14). Instead, it appears that crossl inking has occurred via
Si-C-C-Si, as shown for PSS in Fig. 11-47. An NMR spectrum of PSS-10/
104 shows, in Fig. III-15, a low level of H^C-O-Si and -CH^-OH. This
means that Si(0R)4 not only provides the Si02 phase for the composite,
but also provides crossl inkages for the silane oligomers. Tne TGA char
yields of these composites are given in Fig. III-16; all show higher
yield than silane oligomers without Si(0R)4 and DCP.
An XPS spectrum of PSS-10/924 for Si in Fig. III-17 shows that
there is -20% SiC with respect to Si species in the SiC/Si02 powder com-
posite, as intended. The binding energies of peaks all shifted slightly
towards the high B. E. side due to a differential charging effect. The
B. E. at 102.75 eV should be for SiC and 108.25 eV should be for Si02
and/or Si(0H)4 . In A-PSS-0/927 (Fig. III-13), SiC is shown to be about
one half the Si02 phase. In Fig. III-19, the ViSP/BuOH bulk SiC compos-
ite shows a small amount of SiC and a large amount of surface water or
hydroxyl groups^e of the composite, as represented by a peak at ~536 eV.
Fractured surfaces of SiC/Si02 bulk composites (Figs. III-20 and
III-21) show rougher and coarser textures than a fracture surface of
Si02 gel matrix. There are no differences in surface texture between
80°C dried and 850°C pyrolyzed surfaces of ViSP/BuOH. This is another
151
indication that the composite is still highly porous. The SEM micro-
graph suggests that there is a fracture pattern, as shown by wavy lines,
in the composites seemingly indicating the direction of the fracture.
Microhardness values of SiC impregnated gel composites are much
higher than the gel matrix controls, as shown in Table II 1-3 and Figs.
III-22 and III-23. Unexpectedly, ViSO exhibits a greater hardening
effect than ViSP which has the greater char yield. This may be caused
by two factors: 1) ViSO infiltrated more because of the smaller size of
the molecules, and 2) ViSO infiltrated more easily and stayed inside the
pores because of its higher polarity and, hence, had a higher affinity
to the polar gel matrix by maintaining greater wettability of the gel
surface than the more nonpolar ViSP. As the hardness increased, the
diamond indentation crack became more troublesome in measuring the
indentation sizes, as shown by greater scattering in the DPN values. An
indentation crack for a ViS?/FA gel composite is shown in Fig. III-24.
The strengthening effects of the bulk SiC composite ViSP/BuOH (as
indicated by microhardness) are not as large as the infiltrated compos-
ites (Fig. III-25 and Table III-4). This must be the result of sluggish
densification of the Si02 gel matrix in the presence of dispersed SiC.
Moreover, the hardness value drops sharply with large scattering after
heat treatment above ~800°C, although with little effect on the density
(Figs. III-25, III-26, and III-27). This is attributed to the localized
foaming of the matrix which begins at ~850°C for this type of gel. This
foaming is thought to occur by entrapped gases in the pores.
In hot-stage XRD, an SiC/Si02 composite powder showed Si02 crystal-
line phases at 940°C, but no SiC phase at temperatures up to 14Q0°C.
152
This Si02 phase must come from the oxidation of SiC phase, not from the
crystallization of the SiQ2 glass matrix. 3^ More data on the oxidation
of SiC will be given in the next chapter. The peaks for cristobalite
matured slowly with time at 900°C and did not increase further up to
1400°C in a He atmosphere. This suggests that the SiC phase in the com-
posite hinders the crystallization of SiOo glass; vice versa may well be
true also since no SiC crystalline phase is shown at 1400°C in ~1 hour.
Tnese data are shown in Fig. III-28, along with peaks for the platinum
substrate. A very intense peak at 29 = 39.4° (2.28 A) may be that of
tridymite, but it completely disappeared at 1400°C.
Concl usions
Sol-gel derived monolithic silica glasses can be reinforced by
impregnating with SiC by way of an organosilane precursor. The
reinforcing effect, measured by mi rcohardness, is nearly three times
greater for the SiC infiltrated composite glass after a heat treatment
to 900°C than for the matrix under the same condition. Approximately
100% increase in fracture toughness and a -20% increase in flexural
strength is achieved with the silane impregnation.
Incorporating a SiC phase on a bulk scale can be achieved by mole-
cularly dispersing the SiC from an organosilane precursor in sol-gel
derived silica. For monol ithicity on a large scale, only 2-4 w/o SiC
phase by way of an organosilane is allowed. However, it would be poss-
ible to increase the SiC loading by using a high vapor pressure solvent
in a high temperature mold with an effective sealing capability. This
153
in situ bulk molecular composite gives a reinforcing effect as measured
by an increase in microhardness values. The presumed transformation of
the microstructure and physical properties of the SiC/Si02 composite as
a function of temperature is shown schematically in Fig. III-29.
An obvious problem in all of these composites is in the densifica-
tion procedure. Establishing improvements in the sol -gel processing of
the silica glass matrix is necessary before high performance monolithic
composites can be made routinely.
The experimentally observed "cracker effect" needs to be investi-
gated further to improve composite properties and the fabrication proce-
dure.
A SiC/Si02 molecular composite powder with varying amounts of SiC
phase can be made by mixing an SiC precursor, e.g. an organosilane olig-
omer, and an Si02 precursor. The Si02 precursor SilOR)^ not only pro-
vides the Si02 phase in the composite, but also a crosslinking action as
well. Crystallization of the SiC phase as well as the SiU2 phase is
suppressed in the composites.
Although the objectives of developing a procedure and an under-
standing of the process to produce SiC/Si02 molecular composites have
been achieved, more work to improve properties is needed for actual
applications.
154
T. "C
NecKirii
Liquid
Fig. III-29. Conceptual Microstructure Transformation and Evolution
of Physical Properties of Polysilane Dispersed Sol-Gel
Silica Matrix
CHAPTER IV
SILICON CARBIDE/SILICA COMPOSITES FROM
COMMERCIAL SILICON CARBIDE AND SILICON TETRALKOXIDE
Introduction
A keen interest in high performance materials in recent years has
led to more attention to silicon carbide materials (SiC), as presented
in the previous chapters. However, the excellent mechanical properties
and chemical inertness of SiC have not been fully utilized because of
difficulties in forming and sintering large complex shapes.
Incorporating SiC in a matrix which has desired properties and can
easily be formed into desired shapes and sizes may be an answer to the
problems of forming and sintering SiC. Fiber reinforced composites,
particularly those incorporating SiC as reinforcement are being increas-
ingly utilized for their excellent specific properties, i.e. for high
strength to weight and rigidity to weight ratios.
The reinforcement of brittle materials with high strength fibers
can yield composites of very high toughness. This was first demonstrat-
ed using carbon fibers in glass and glass-ceramics. 88"90 More recently
the availability of continuous SiC fiber has led to tougher glass and
glass-ceramic composites^^'^"* which are more resistant to high tempera-
ture oxidation than the carbon fiber composites.
Despite its remarkable mechanical properties and chemical stability
in ambient conditions, gradual oxidation at high temperature limits the
155
156
wider applications of SiC as a truly high performance/hiyti temperature
material .
Tnere are several mechanisms known for the oxidation of SiC in the
temperature range of 1000°-150U°C.95-ioi /\t low oxygen partial pressure
(Pq ), < ~3 X 10"^ atm at 1400°C, an "active" oxidation occurs due to
the formation of gaseous products as shown by the Eqs. IV-1 and IV-2
below. s**
SiC(S) + 02(g) t SiO(g) + CO(g) (IV-1)
SiC(S) + 3/202(g) t SiO(g) + C02(g) (IV-2)
At high Pg , "passive" oxidation due to a protective film of Si02 is
operative according to the reaction Eqs. IV-3 and IV-4.100
2 SiC(S) + 302(g) * 2Si02(S) + 2C0(y) (IV-3)
SiC(S) + 202(g) ^ Si02(S) + C02(g) (IV-4)
At any case, a gaseous product or products are formed.
If the latter mechanism of passive Si02 film formation is applic-
able for SiC oxidation, an intentional coating of the surface of SiC
material with Si02 glass should help to prevent or passivata the further
oxidation of SiC. Moreover, the Si02 glass matrix may be used to im-
prove fabricabil ity of SiC by viscous deformation at much lower tempera-
tures. Thus the incorporation of Si02 glass into a SiC skeletal struc-
ture or vice versa by forming a composite has three potential advan-
tages, 1) improving the fabricabil ity of SiC, 2) reinforcing the matrix,
and 3) minimizing the oxidation of SiC.
In spite of these obvious potential advantages, there have been
only a few studies to investigate how a Si02 matrix can affect the oxi-
dation kinetics of a SiC/Si02 composite at high temperatures.
157
The forming method used by the previous investigators ''5-78 -jri
making SiC/SiOo composites was the conventional hot pressing technique
which has great limitations on sizes, shapes and complexities of the
composite bodies. It is hypothesized that these limitations nay be
overcome by way of sol-gel processing. As discussed in the previous
chapters, one of advantages of the sol-gel method is in the simplicity
and ease of forming a green body. By casting the sol into inexpensive
or disposable plastic molds followed by aging in an oven, a monolithic
green body can be produced in almost any shape and size. The previous
chapters describe methods for preparing a very low volune fraction of
SiC in a Si02 matrix to form composites. In this chapter, however, a
process is decribed to achieve a much greater SiC loading in the Si02
matrix derived from sol-gel technique and commercially available SiC
fibers and whiskers.
A lengthy review on ceramic-matrix composites was reported by
Donald and McMillan. 59 Factors in designing and making fiber-ceramic
composites were given by Bialoskorski and Konsztowicz. ^^2 jhg factors
include lengths of fibers, their contents, and types of fibers used for
reinforcement. Wang and Sutula^^s showed that the detrimental effect
due to the difference in thermal expansion between fiber and metal mat-
rix in metal -matrix composites is minimized for short fibers. Lannutti
and Clarki0'+"i06 showed a potential usefulness of sol-gel derived alum-
ina in SiC/Al203 composites using whiskers, mats, weaves, and short
fibers of SiC. Rice et al.if^ showed the effect of interfacial bond
strength between the fiber and the matrix on mechanical properties.
158
Some of important factors affecting mechanical properties of fiber
reinforced ceramic composites are listed below.
1. Volume Fractions of Components
2. Porosity of the Composite
Lower volume fraction of continuous phase often
increases porosity
3. Ultimate Strength of Fiber and Matrix
4. Interfacial Bonding
High bond strength between fiber and matrix
Low bond strength between fiber and matrix
5. Thermal Expansion Coefficient Match
Comparability of a thermal property at high temperature
6. Flaws, Defects, Impurities, etc.
The objectives of the work in this chapter are 1) developing a pro-
cedure to produce SiC/Si02 composites using the sol-gel process, and 2)
understanding the variables affecting mechanical properties.
Experimental
Composites of 6-SiC and a-SiC in a pure silica gel matrix were pre-
pared using Nicalon® and Silar" as the reinforcing filler phase.
Chopped fibers of Nicalon were pretreated to remove a polyvinyl
acetate coating by successive washing in ethyl acetate, benzene, and
* 6-SiC by Nippon Carbon Co., distributed through Dow Corning, Midland,
MI.
**a-SiC by ARCO, Greer, SC.
159
acetone followed by firing at 400°C in air for two hours. The cleaned
Nicalon was further chopped in a polypropylene container with alumina
balls on a vibratory mill for one hour. In addition, continous fibers
and weaves of Nicalon were cut into the sizes of molds and the polymer
coating was removed by burning it on a propane burner. Silar" was used
in its as-received form or after heating in air at 300°C for one hour.
Scanning electron micrographs of Silar™ and the chopped Nicalon® are
shown in Fig. IV-1.
Characteristics of the SiC used are listed in Table IV-1.
Silica sol was prepared by hydrolyzing tetraethyoxysil ane (TEOS)
with HCl in ethanol as the solvent in the mole ratio of 1:4:0.5 for
TEOS:water:alcohol . Ten grams of SiC was mixed with 75 ml of the silica
sol containing 1-5 ml of glycerol and/or 1-5 ml of formamide as drying
control chemical additives (OCCA's) for 1-30 min followed by ultrasoni-
cation before casting in polystyrene molds of various shapes and sizes.
The various configurations of the composites made in this manner are
shown in Figs. IV-2 and IV-3.
The SiC/Si02 sol slurries, after being tightly sealed, were aged in
an oven at 40-80°C for ~10 hours before slowly drying in air for 5 hours
at 90°C. The dried green composite bodies were impregnated with Si Go
sol up to four impregnation-drying cycles followed by a vacuum impregna-
tion.
Cylindrical composite bodies for oxidation experiments were cut in-
to thin wafers using a diamond rotating blade. The wafers were dipped
into the silica sol twice after each drying in oven. The composite
160
®
Fig.IV-1. SEM Micrographs of Chopped Nicalon Fiber (top) and
Silar^ Whisker (bottom)
151
Table IV-1. Properties of SiC Used in SiC/Si02 Composites,
Data Provided by the Manufacturers
NICALON®
SILAR scg"
Tensile Strength - MPa
2000-2520
689U
Tensile Modulus - GPa
180-200
689
Elongation - %
1. 5-2.0
—
Density - g/cc
2.55
3.17
Filament Diameter - ym
13-15
0.6
Length - ym
Infinite
10-80
Cross Section
Round
Hexagonal
Type of SiC
Beta
Alpha
Maximum Temperature
1200°C
1760°C
Form
Chopped or
Whisker
Continuous
Single Crystal 1 ine
Fiber
Surface Area - m^/g
Impurities - w/o
-20 Oxide
-IS Free
Graphite
Trace of Metals and
< 3% Oxide
162
SiC/SiQ
SLURRY
SiC CONTINUOUS
FIBER / S1O2
SiC WEAVE/
Si Op LAYER
SiC CHOPPED fiber/
Si O2 LAYER '
Fig.iy-2. Some of Layouts Used in Fabrication of Nicalon/SiO,
Gel Matrix Composite ^
153
SiC/Sol-Gel SiO 2 Composites
Nicalon ®
Silar
TM
Ocm1 Z 3-4"5
Fig.IV-3. Cast Nicalon /Si02 Gel and Silar"" /5i02 Gel
Composites of Various Shapes and Sizes.
164
wafers were exposed to 1100°C in a box furnace with dry static air and
sequential oxidation of the wafers were carried out. The samples were
heated to 1100°C beginning from room temperature each time at a rate of
~200°C/hr.
The wafers contained ~55% SiC, -26% Si02, ~15% open pores, and --6%
closed pores by volume, on the average. The open porosity was measured
by mercury porosimetry and the closed pores estimated by the theoretical
density and a subtraction method. Excellent dimensional stability of
the composite sample wafers was maintained. There was no warpage and
the shrinkage was < 5 v/o. The mean density was 1.67 g/cm^ which is
~70% of the theoretical density of the composite wafers. The density
change was + 0.2 g/cc during the oxidation.
After each oxidation exposure of the sample, FT-IR spectra were
taken by using a Nicolet MX-1 FT-IR spectrophotometer in diffuse reflec-
tance mode. Thermogravimetric analysis in Pq =1 atm. was carried out
in continuous flowing dry O2 at 1000°C using a DuPont TGA model 1090.
Some of composites were made using ~ one-third the sol used in the
cast composite above followed by cold pressing in a steel die (Carver
Laboratory Press) at -10,000 lbs load for 5 min at 25-80°C followed by
oven drying and a vacuum impregnation of Si02 sol. The steel die was
premachined to form notched sample bars for fracture toughness as shown
in Fig. IV-4 as well as unnotched bars for a flexural strength test.
Three point flexural strengths were determined by using an Instron Test-
ing machine and the compressive modulus was measured with an MTS mach-
ine. A three-point instead of four-point test was used because of the
specimen sizes. Microhardness was measured with a Kentron Tester using
Ram
Mold
Base
for C,
flex
Fig.IV-4. Steel Mold Used to Cold Press
Base
for K,c
Nicalon®/Si02 Gel
Silar"'"'^/SiOp Gel Composite Specimen for<^fi
ex
and
and Kj(3
166
1-2 kg loads. Porosity was measured usiny mercury jjorosimetry and an
Autosorb-6. The composite bodies were measured by mercury volume dis-
placement after each heat treatment. Measurements of fracture toughness
(Kj(-) were obtained by using the conventional 3-point bend tests on
notched beams same way as for the 3-point flexural strengths. The in-
dentation method^s described in Chapter III was also used to determine
KjQ and compared with the notched beam 3-point bending test.
A test for thermal shock resistance was carried out by heating cold
pressed composite bar samples to 800°C in N2 ^'^'^ quenching them in a
silicone oil bath followed by flexural strength measurements. Effect of
oxidation on strength of the composites was examined by heating the
specimen in air at 90U°C for four hours followed by flexural strength
measurements.
Transmission electron microscopy (TEM) of Nicalon® and Silar" com-
posites was used to examine the interfacial region between the carbide
and the oxide. To make the TEM specimen, thin disk composites were
formed and cut into ~3 mm diameter disks, then reduced down to ~1 mm
thickness by an ultrasonic saw. The disks were further polished down to
~150 urn with successive Carborundum polishing steps. The -150 ym thick
disks were dimpled by a VCR dimpler. At this stage, the middle portion
of the samples was ~30-50 ym. Further thinning until a perforation in
the middle was obtained by using an Ar ion beam of 100 uA current and 6
KV potential. It typically took ~15-20 hours for the ion milling.
Transmission electron micrographs were taken using a JEOL, JEM-200CX
* VCR Group, San Francisco, CA.
167
Analytical Electron Microscope with a 20 nm beam size and ~10U sec
counting times. Thermomechanical analysis (TMA) and differential scan-
ning calorimetry (DSC) analyses were obtained with a DuPont 1090 Thermal
Analyzer.
The effect of OS impregnation into the porous NC and SC was examin-
ed by soaking the dried specimens into benzene or THF solution (~2 g
OS/IO ml solvent) for 2-5 hours followed by pyrolysis at 900°C for 2-4
hours in No and Of-^Q^ measurements with an Instron Testing machine and
KjQ measurements with a Leco Microhardness Tester.
Results
The SEM micrographs of Nicalon® and Silar" in Fig. IV-1 show a ~10
ym diameter for Nicalon®, and ~1 pm for Silar", and a mean length of 30
ym for both types of fibers. This gives R - 3 and R ~ 30 for Nicalon*^
and Silar", respectively.
Differential scanning calorimetry of the as-received Nicalon® is
shown in Fig. IV-5. The exothermic peak at ~320°C at the first heating
in air is a result of the oxidation of the polyvinyl acetate coating on
the fiber used for sizing. The peak for oxidation of the coating is ab-
sent in the second heating. The TGA in Fig. IV-6 agrees well with the
DSC. A significant weight gain by the oxidation of SiC Nicalon® at
temperatures above ~750°C is shown in Fig. IV-6. Thermomechanical anal-
yses (TMA) of a Nicalon composite (NC) and a Silar composite (SC) after
various heat treatments are shown in Figs. IV-7 and IV-3. Negative ex-
pansions caused by sintering are shown for all cases.
168
100
Temp^<>C
400
®
600
700
Fig.IV-5. DSC Thermograms of as-received Nicalon Heating in Air Twice,
— Showing the Removal
The First Heating and second —
of the Sizing Polyvinylacetate at ~320°C
69
IV ' '. ■ ^.'.V ' ■"• ■■TTrtAT.-i^G
>^
J2
c:
-t— '
—
o
CTl
en
■ ^
o
i4-)
-^
en
^t—
• r—
o
3
to
i/i
4->
o
c
1
03
t^
CTl
• .—
C
c
■ -—
CTl
3
• f—
O
oo
r~
UO
i_
o
CJ
O
XI
CNJ
^
■ .—
r^.
3
I—
;
«c
■^
v^
o
rz
f^
ui
ro
cn
•r—
CJ
c
•1—
■ f—
'^
T'
NJ
•^-
O
-a
LO
o
CJ
CO
>
<U
(
•^
■^
OJ
ra
QJ
o
4-J
>
OJ
Ol
o
1_
u
-D
1
f^
<
lyi
r—
ra
>>
^
o
tj_
■ —
o
>
• -—
>,
u.
un
^—
Zl
o
<_)
-0
c_
■r—
i.
^J-l
cn
Ol
o
_c
o
+^
C
-4->
o
l+-
_c
o
M-
1 —
c
o
<:
o
c
o
o
h-
4~>
Q
o
+J
-o
03
■o
■*—
-o
C
X
■f—
rn
o
o
cC
a
o
>>
1 —
■M
j::
170
<u
c
o
to
en
s-
o
cc
o
(/) . — .
O CJ
I— QJ
QJ +J
C3 +->
O
o —
•r- S_
®"^
o s_
r— O
U
C
■!-> C
t/i a^
03
c_> 4-J
o:
U-
O en
c
on T-
E ■•-'
to ra
S- QJ
cnr:
o
£ s-
s- QJ
CI +->
I— «=:
cC -a
I— (O
I
171
o
00
-t->
C7)
!=
>>
S-
o
s-
<u
■*->
(U
(/I
o
Q.
B
o
<_3
C3'
CM •
O S-
r
h
i-
i-
o
ra
M-
r^
■r—
t_)
oo
O
+->
O
1/5
C-.
ra
<_>
+->
fO
M-
O
CD
£=
CO
■ r-
E
4->
<a
fd
s-
(U
D13:
o
E
s_
s-
<u
<u
■(->
JZ
<4-
h-
<
< -o
s:
c
I—
(O
CO
1
o
ujrf*afi&vt|0 'ui|a
o
I
o
I
o>
172
The compressive modulus of cast NC after a 1,00U°C heat treatment,
yielding a density of 1.79 y/cc, is 571 MPa. Flexural streng-,hs (af^g^)'
densities, and microhardnesses of the composites are given in Table
IV-1. Table IV-2 lists the BET surface areas, pore sizes, and the total
pore volume of NC and SC. Densities as a function of infiltration cycle
and a function of heat treatment temperature are yiven in Table IV-3.
Density and afigj. as a function of processing temperature for 1UU%
chopped fiber NC is given in Table IV-4. The maximum a^^g^ of NC with
different layouts and processing conditions after 1100°C heat treatments
for two hours are given in Table IV-5.
Graphical representations of density vs. firing temperature for
cast NC and SC are shown in Fig. IV-9. Microhardness (DPN) vs. density
for cold pressed NC and SC are shown in Fig. IV-10. Greater density and
DPN are shown for NC than SC. Flexural strength vs. DPN in Fig. IV-11
also show an exponential type of function. '^f^Qx ^^* ^l^^sity in Fig.
IV-12 shows a straight line function for both NC and SC. The pressed
Silar composites (SC) have nearly 2-3x the a-f-jg^ of pressed NC at the
same density.
A sample specimen for the fracture toughness represented by Kjq is
shown in Fig. IV-13. The parameters to calculate Kjq and the resulting
KjQ values for cold pressed NC and SC are given in Table IV-5. Typical
load vs. crosshead displacement curves are shown in Figs. IV-14, IV-lb,
and IV-16.
Calculation of flexural strengths was done according to Eqn.
IV-5. 108
173
Table IV-1. Three-Point Flexural Strengths, Densities, and
Microhardness and the Ratio of Strength to Density of NC and SC.
Specimen a^lg^jMPa p, g/cc Hardness, DPN c^flex'^P
Cast NC
100% chopped
fiber, 800°C
19+2 (2)
1.64±0.1
370±25
11.6
100% chopped
fiber, 140U°C
83±3 (2)
1.82±0.2
660±37
45.6
100% chopped
fiber, ViSP
infilt., 800°C
37.2±3 (2)
1.90±0.1
467±21
19.5
100% chopped
fiber, ViSP
infilt., 800°C
33+1 (2)
1.23±0.1
520±52
26.8
unidi rectional
fiber, J-PSSl
infilt., 1000°C
9.3+4 (2)
1.66±0.1
55U48
5.6
unidi rectional
fiber, 140U°C
53+6 (2)
12U±5 (2)
1.82±0.1
1.98+0.1
1051±55
29.1
60.6
Pressed NC
80°C dried
12±1 (2)
1.73±0.2
6.9
J-PSSl infilt.
800°C
31±2 (2)
1.85±U.l
16.8
800°C and
quenched to RT
115±5 (2)
22±1 (2)
2.07±0.1
1.72±0.1
915±58
55.5
900°C
1100°C
1400°C
19±5 (2)
45±3 (2)
47+4 (2)
1.92+0.1
2.18+0.1
958±63
23.4
21.6
900°C, 4 hrs
in Ng
25 (3)
29
40
1.90
2.01
1.97
13
14
20
900°C, 4 hrs
in air
23 (3)
22
35
2.00
1.88
1.65
12
12
21
174
Table IV-1 (continued).
Specimen a^-]g^,M°a p, g/cc Hardness, DPN °f]ex'°
Cast SC
1000°C
10±2
(3)
1.45±0.2
378±76
6.9
Pressed SC
80°C dried
18±3
(2)
1.41±0.1
12.8
700°C
31±2
(2)
1.80±0.1
17.2
800°C
42±3
(2)
900°C
88±6
(2)
1.87±0.1
47
950°C
116±£
i (2)
1.88±0.1
62
ViSP infilt.
820°C
53+3
(2)
1.86+0.1
403
28.5
900°C
89+7
(3)
1.91+0.1
413+42
46.6
950°C
216+11 (2)
1.83+0.1
1284±85
118.0
ViSP infilt.
950°C
78+6
(2)
1.94+0.1
40.2
1400°C
112±8 (2)
2.08+0.1
53.8
800°C quenched
i25±;
' (2)
1.82±0.1
805±25
68.7
to RT
900°C in N2
4 hrs
39
1.92±0.1
20
149
1.72±0.1
87
117
1.76±0.1
66
88
1.71±0.1
51
68
1.87±0.1
36
900°C in air
45
1.88±0.1
24
4 hrs
101
1.91+0.1
53
95
1.75±0.1
54
The numbers in parentheses denote the number of specimens tested.
175
Table IV-2. The BET Surface Areas, Mean Pore Sizes,
and Total Pore Volumes of NC and SC.
Heat T
rec
itment
BET
Me.
an Pore
Total Pore
Composite
Temp
• >
°C
Area, m^
Vl
S
ize, A
Vol . , cc/g
Cast NC
300
in
ai r
141
16.2
0.114
800
in
N2
45
16
0.036
1500
ir
1 ai r
0.86
163
0.0071
ViSP infilt.
800
in
Ng
99
13
0.063
NC
b.
Pressed NC
300
900
73
4.6
14
68
0.0517
0.0155
Cast SC
300
850
184
18
17
14
0.16
0.13
Pressed SC
300
850
143
5.8
18
140
0.12
0.040
Pressed/
800
29
33
0.049
ViSP infilt.
SC
176
Table IV-3. Density Changes as Function of Processing
Temperature and Sol Impregnation Cycles for Cast NC and SC,
Dipping Cycle After
500°C for 2 Hours
1
2
3
4
5
Density, g/cc ±0.1
NC
SC
1.44
1.11
1.56
1.29
1.70
1.41
1.81
1.52
1.86
1.59
1.87
1.62
Temperature, °C
After 2 Dipping Cycles
500
900
950
1100
1400
Density, g/cc ±0.1
NC
SC
1.72
1.41
1.80
1.49
1.88
1.54
1.98
1.62
2.11
1.73
Two specimens were used for each cycle and each temperature.
177
Table IV-4. Density and Flexural Strengths of Cast
NC 100% Chopped Fiber as Function of Processing Temperature.
Temperature, °C
p, g/cc ± 0.1
Gfiex' ^'P^
^flex
1.52
15±4
9.26
1.71
17±3
9.94
1.83
20±4
10.9
1.95
27±5
13.8
2.06
83+3
40.3
80
500
900
1100
1400
Two specimens were tested for each tenperature.
178
Table IV-5. Maxinum a^ig^ °^ ^^ °^ Different
Layouts and after Processing at 1100°C or 1400°C,
Composite "^^flex' ^^^^
cast 100% 27±5 (3)
chopped fiber 83±3 (1400°C) (2)
bidirectional weave/ 15±3 (4)
chopped fiber
continuous fiber/ 25+4 (2)
chopped fiber 120±5 (1400°C) (2)
cold pressed 115±12 (3'
100% chopped fiber
The numbers in parentheses denote the numbers of specimens tested.
179
2.4 ^
Fig.IV-9. Density as a Function of Firing Temperature of
Cast Nicalon®/Si02 Gel Composites and Silar"""/
SiO„ Gel Composites
180
1200 -
1000 -
1.4
t
2.0
1.6 P :^A^^-9
Fig.IV-lO.Microhardness as aFunction of Density for Cast Nicalon /
SiOp Gel and Silar'^^/SiOj Gel Composites
181
1400
1200
^
1000-
800
Z
600
400
200
20
40
60
C^lex ,A^P
L.
00
120 14(5 ^^ rio
Fig.IV-n. Microhardness vs Flexural Strenth of Cold Pressed
NC and SC
182
2.2
f. 3/c.
-®/
Fig.IV-12. Flexural Strengths of Pressed Nicalon /SiO Gel and
Silar"'"/Si02 Gel Composites As a Function of Density
183
Molded Notch
4.4Cm
Fig.IV-13. Schematic of Notch-Beam Test in Three-Point Bending for
Fracture Toughness of Pressed NC and SC
to
<u
3
o
184
1^
a.
3
1+
c\j
o
CM
1+
30
"C ^
01
o
i.i:
+J ;:^
2-
01 ^
O
00
t/1
o
J3
o
O
0)
a;
CM
CO
■o
o
CO
CM
Lfi
CM
CO
CO
•^T,
r— 1
un
o^
r^
o
a
iX!
CM
Parameters
Cold Presse
>-
•
•
«
•
CM
CM
CM
CM
CM
s
CO
CO
LP
=3-
■^^
•
•
•
•
ro
o
o
o
O
, ^^-
LO
LO
1^
"^
1
o
o
•
•
•
o
•
>
o
o
o
o
C71
+1
+1
+1
+1
1 — 1
1 — 1
<o
CO
<U
#■
C7\
o
00
o^
a
•
•
•
•
<-i
i-H
CM
«-H
.— 1
lO
1—
CJ
o
ID
o
c_;
o
4->
o
o
o
• r—
o
o
O
o
1/1
o
^
o
^
o
Ln
f— 4
LO
I— 1
Q.
E
*t
M
M
M
O
o
o
CJ!
C_)
o
^
2:
00
OO
T3
<U
(/5
(U
{_
<u
3
OJ
•I—
o
E
o
o
(J
o
o
5
185
20
15
10
Cast, 1100°C
Nicalon® Continuous fib
_L
J.
0.05
0.1
0.3
Fig
0.15 0.2 0.25
Crosshead Disp. , inch
IV-14. Load vs Crosshead Displacement of ViSP Impregnated^ SC A/ter^
Pyrolysis at
900
*C{top) and Micalorf Cont.. Fiber/Si02 Gel Comp .( bottom)
186
sqx 'peal
187
sq-[ ' pecq
188
0,1,, = 3/2 ^^Q^^^-^P^") ^ (IV-b)
^^ (width) (thickness)^
Calculation of Kjq is based on Bansal and Duckworth ^o^ and shown by Equ.
IV-6.
1/2
K,- = ^ ^^^,. ^ Y (IV-6)
^^ b Vl^
where M = applied bending moment at fracture = P x S
Y = dimensionless paraneter which depends on a/w and type of
loading
= Aq + A^ (a/w) + A2(a/w)2 + A3(a/w)3 + A4(a/w)'^
for s/w = 8, Aq = 1.96, A^ = -2.73, A2 = 13.66, A3 = -23.98, and A4 =
25.22.
A fracture toughness measurement by a Vicker indentation method^3
has been described in Chapter III. The results of the crack lengths,
hardness, and Kjq are summarized in Table IV-7. The relation between
crack length and Kjq is as follows, although the validity of this method
has not yet been fully proven for composite materials.
H = ^—y and K.p = 0.016 (|)^/^ ^^ (IV-1)
2(a/2)^ ^^ " (c/2)-^^'^
where: H = hardness in Pa
P = load in newtons
a= mean diameter of Vicker indentation
c= mean diameter of the extended crack due to loading
E = elastic modulus estimated using the rule of mixture
The TEM micrographs of cast NC and SC before and after heat treat-
ment at 900°C are shown in Figs. IV-17, IV-18, IV-19, and IV-20.
189
o
o
o
00
■a
(T3
■o
c
_i
fO
cr>
o
ji<;
2:
1— 1
-o
(U
t_
(/I
0)
(JO
■o
a;
c
t_
rD
a.
i/i
"O
^
4J
"o
O)
'-_)
c
0)
<*-
_i
.^.^
to
CO
fO
<D
i_
C
CJ
JZ
CD
c
3
•r-
1—
4J
(O
a;
■(->
c-
c
3
0)
-M
■a
c
ra
»— *
c_
Ll-
>.
^
a.
-H
-H
O
>£)
C\J CM
r— I CM
• •
(O
t — 1
t — 1
Q.
CM
CM
-H
-H
ts
-H
■H
CO
CM
CM
LT)
•
•
«t
.— I
.—1
LO
U3
3.
10
-H -H
00 ir>
vo
I —
CO
3.
-H
-H
-H
-H
(Ti
CO
r~~
•>
"^
00
LD
LT)
ItJ
3
I.
o
0)
ex
CO
o
1 c
n3
C3
CM
CM
CM
CM
^
>*
'a-
>*
UJ
CM
CM
CM
CM
^ Ol
03 Q
.—1
)— 1
»-H
f-H
•
•
•
•
-^.
-H
-H
-H
-H
en
VO
1^
i-H
CM
<J\
CTl
A
•
•
•
•
a
i—t
t— 1
CM
C\J
<_3
•■-
«J)
-1-
30
20
e
• a^
• 1^
0)
"^
^ 4J
«3-
•a- +->
r—
f— 4
1 — 1 1 — •*
i-H
1 — 1 1 — «
a.
•.- Q.
•.- Q-
E
•>
••>«- CO
«s
-M- 00
(O
C_) C -i-
CJ
C -1-
00
z
Z r- >
00
00 -1- =>
■)->
c
(U
C-
3
(/)
<u
XJ
c
10
CO
(U
O)
a.
CO
ro
I
CM
190
Fig.IV-17. Bright Field TEM Micrograph of NC After 80°C Drying
at 59KX Mag. (top) and STEM Micrograph of the Above
at 300X Mag. (bottom)
191
Fig.IV-18. Bright Field TEM Micrographs of Cast
Heat Treatment in N^. Top: 50KX Mag.
NC After 900 C
Bottom: 1 KX Mag,
Darker Contrast = fiber and Lighter Cont. = Silica
192
Fig.IV-19. Bright Field TEM Micrographs of SC After Drying at
80°C, Top: 50KX Mag. and Bottom: 200KX Mag.
193
Fig.IV-20.
Bright
in No.
Field TEM Micrographs of SC After Heating at 900'
Top: 50KX Mag. and Bottom: 200KX Mag..
194
Figure IV-21 shows SEM micrographs of a surface of cast NC and SC
after drying at 80°C. Figure IV-ZZ shows a polished surface of NC after
1100°C for two hours in N2. Figure IV-23 shows fractured surfaces of
unidirectional continuous Nicalon® fiber in Si02 matrix showing dislodg-
ing of the interfaces and of random 100% chopped fiber NC.
The extent of 1100°C dry static air oxidation of SiC in the Si02
composite was quantitatively estimated using reflectance values (R) of
FT-IR and the base-line method. -i*' The base-line method was applied to
compensate for the fluctuation in IR response of the instrument. A few
typical FT-IR spectra are given in Figs. IV-24, IV-25, IV-25, IV-27, and
IV-28. The IR bands to estimate R values of each species are tabulated
in Table IV-8. Reflectances at -789 cm"^ (R^.), -1086 cm'^ (R ), and
-1157 cm"-'- (R^) are due to vibrational motions of the Si-C bond,
vitreous Si-0 bond, and cristobalite Si-0, respecti vely . ^ ^ ^"i 13 Plots
of time vs. R values were made and are shown in Figs. IV-29, IV-3U,
IV-31, and IV-32.
In the composites, the total amount of vitreous silica is shown to
be relatively constant throughout the oxidative heat treatments. This
means that the oxidation of SiC in Si02 glass produces predominantly a
crystalline Si02, namely, cristobalite. This was demonstrated by Figs.
IV-24, IV-25, IV-26, IV-27, and IV-28.
Thus the normalization of R's by R„ should compensate R due to sam-
ple shapes, sizes, and surface morphology after each heat treatment.
The results of isothermal TGA carried out in dry static oxygen are
given in Figs. IV-33, IV-34, IV-35, and IV-36. The weight gains are
tabulated and given in Table IV-9.
195
Fig.IV-21. SEM Micrographs of As-Cast NC (top) and As-Cast SC
(bottom) After Drying at 80°C.
196
Fig.IV-22. SEM Micrographs of Polished Surface of NC After 1100°C
Heating in N2.
197
Fig.IV-23. Fractographs of NC, Top: Uniaxial fiber after 1C00°C
Bottom: Random chopped fibers after 1100''C
198
O
o
30NV103nd3a %
o
<:
o
c
(£)
en
O
O
00
4-1
<
"f
QJ
o
O
8
a:
IxJ
Q-
o
O
GQ
CNj
^
O
Z)
Ut
2
LU
c
o
§
o
o
^
u
CO
•'—
O
E
3
o
o
^
Q.
Ci
1
f—
Ll.
o
.
o
OJ
a>
1
199
CO
S-
-a
f3
o
S-
30NV103nd3a 7o
o
o
o
o
O
o
en
(U
CJ
3
o
Q.
to
c::
t— t
I
I—
ID
CM
I
200
30NV103nj3d 7o
201
33NV103nJ3a 7c
u.
202
o
o
o
o
lO
o
■!->
03
o
<
IT)
o
r-
•r-
+->
(/I
c
o
en
o
^^
c
o
T
4->
£
■*--»
C/)
4-
o
cr
o
iij
GQ
2
00
3
CM
LlJ
>
<
14_
<
O
^
O
CO
in
4-
O
E
3
S_
+->
U
O
a.
(/I
h-
30NV103nd3d %
o
o
o
CM
CO
CSJ
203
Table IV-8. Summary of FT-IR Spectra for a Sequential
Oxidation of SiC and SiC/Sol-Gel Derived SiOo Conposite in
Dry Static Air at 1100°C, ~55 v/o SiC, -25 v/o SiO^.
Exposure
~"l " _ 1 _
Time, hr "^x, cm"-^ R„ ^'g
A
, cm"-'-
«y
^c, cm-1
«c
R,/Rg
R,/Ry
Nicalon® (s-SiC) Composite
1167 0.004
1086
0.10
789
U.072
0.04
0.72
5 1257 0.22
1114
0.15
800
0.14
1.44
1.0
10 1260 0.14
1118
0.30
798
0.14
0.47
0.53
16 1265 0.49
1120
0.31
794
0.10
2.0
0.78
Raw Nicalon® (s-SiC) Parti ci
jlate
undetectable
790
0.1
oo
11 1256 0.075
1104
0.16
798
0.069
0.47
0.43
26 1260 0.15
1112
0.17
816
0.10
0.88
0.59
Silar" (a-SiC) Composite
1126
0.01
1087
0.13
792
0.15
0.08
1.16
5
1246
0.17
1126
0.15
790
0.20
1.13
1.37
10
1250
0.34
1124
0.22
788
0.38
1.55
1.72
16
1256
0.27
1126
0.14
788
0.21
1.93
1.50
26
1260
0.31
1126
0.11
736
0.17
2.82
1.55
Raw Silar" (a-SiC) Whisker
1259
0.029
1152
0.038
805
0.293
>0.77
>7.7
11
1282
0.054
1156
0.015
806
0.423
>3.6
28
26
1248
0.142
1144
<0.01
764
0.249
>14
>25
204
2.4
O Rx/Rg
ARc/Rg
NC
10 15 20 30
Time, hr
40
Fig.r/-29. FT-IR Rflectance of NC As a Function of the Exposure
Time in Air at nOO°C for a Carbide and a Cristobal ite
205
10 15 20
Time, hr
Fig.IV-30. FT-IR Reflectance of Raw Nicalon As a Function of Time
Under Oxidative Exposure at 1100°C in Air
206
X X
Composite <^ R^ A^
0.5 L S^9 O
Raw Siiari'^ r^^ p
SC
10 15 20
Time, hr
25
30
Fig.IV-31. FT-IR Reflectance of Silar'7Si02 Composite and
Raw Silar^'^As a Function of Time Under Oxidative
Exposure
207
R
10 15
Time, hr
20
25
Fig.IV-32. FT-IR Reflectance of Silar /SiOo Composite and Raw Sil
As a Function of Time Under Oxidative Exposure, After
Normal ization
208
3
o
o
O
CO
o
00
o
B
c
T
c
o
o
o
-i-j
fe^.
(t3
C\J
L E
o
>1
'° -."
c
E
c
o
o
s-
Ol
■>->
QJ
o
•1-)
l/l
(B
t— 1
Di
CO
1
209
o
o
7
o
o
o
CM
c
C C2
o @
.! §
3
cc:
o
=1
C3
ns 0)
E +->
S_ (13
o o
CO
210
c
E
E
O
o
to
E
o
T
o
c
o
c
1/1
o
O)
o
t/1
CO
I
9^ '•4M
211
o
o
o
o
o
o
o
00
a:
3
O
u
o
T
<_3
o
o
o
o
c-
o
CM
o
O
Vi)
>>
C
s_
o
o
o
E
n
«.
<D
r
£
r
o
o
ra
1—
^
tr
3
o
o
<+-
rr
O
<
o
o
1/1
I
t.
4M
212
o
r — •
a>
-\
.
+->
c
3
2"
QJ
•_'
^—
a.
<
c:
(O
oo
VO
^
a>
CM
'^
'S-
n
iD
C3^
-:vi
fo
=0
en
o
o
^C
f— t
CO
. — t
.— I
>1
O
O
o
4-1
., ,
a>
■ — '
*j
fO
r—
<D
+->
r^
f—
^
c
c
►— 1
TJ
OO
r--
( — 1
O
O
O
,^
1 — 1
. — 1
CM
CM
cc
cc
CO
CT^
ZT\
CM
CM
CM
ro
CO
o
o
o
CO
n
o
o
f~^
o
o
c: o
CO
O
Q.
E
o
CM
O
OO
<_)
(^
CM
CM
LO
•X)
ID
00
•
•
CM
CM
cn
LO
c
o
I.
o
CM
O
CM
o
•
r^
^
CO
C^J
ro
o^
(— (
' — '
ro
<:
i_
o
■r- +J
C 3
CM
CM
•
•
o
.-H
T-H
o
o
•
CM
CM
CM
•
•
o
o
CO
o
o
O
o
o
.— 1
1-H
cn
* — 1
1—1
r-H
a>
3^
I
(U
3
(/) •
O 0)
Q. E
X -1-
ir>
c
o
^—
«
o
(U
T-
21
"q.
E
3
«
<o
CO
oc
CM
o
• r—
(/)
t_
^^
IT3
c
^.-
o
•^-
t/l
TJ
O
5
<o
z
a:
CM
c
CO
CO
213
A heat treatment of the composites up to 1500°C in air showed no
visible change in shape or physical state. Hov;ever, heating the compos-
ite bodies in vacuum (~10"^ torn) at 1400°C resulted in a disintegration
of the composite body and an evolved gas resembling H2S. This must have
been caused by some type of chemcial reactions between the two phases
which has been noted in Chapter III.
Discussion
Several differences between the fibrous SiC Nicalon® and Silar" can
be listed:
1. phase: Nicalon®; 6-SiC
Silar"; a-SiC
2. dimension: Nicalon® is ~10 times greater in
diameter and length
3. crystalline form: Nicalon® is polycrystall ine fiber
Silar™ is single crystal whisker
Extremely large particle sized SiC phases i.e. continuous fibers
and weaves as shown in Fig. IV-2 showed much lower density and strength
in cast NC. Fracture energy, however, is expected to be higher as shown
in Fig. IV-16.
The polyvinyl acetate sizing on Nicalon® could be removed by heat-
ing in air at ~35U°C as shown in Figs. IV-5 and IV-6. As shown in Fig.
I\/-6, heating the fiber in air above ~700°C caused weight gain by SiOn
formation.
Shrinkage due to heating begins at ~200°C and accelerated above
~700°C (Figs. IV-7 and IV-8). At 700°C -10 m linear shrinkage is shown
214
for both NC and SC. Greater slope for SC after drying at 80°C is shown.
Considering the initial sample sizes, NC shows a greater densification
rate of 0.48% linear shrinkage vs. 0.26% linear shrinkage of SC in ten
hours. This must be partly due to the greater initial density of NC in-
dicative of better compaction of Nicalon® and suggests a better compati-
bility between the silica sol and the Nicalon® fiber phase than that be-
tween the Silar whiskers and the matrix. This agrees with the results
in Fig. IV-9.
The three-point flexural strengths given in Table IV-1 indicate
that there is general agreement among of-^Qy^, p, and hardness. However
the large variations in Of-^Q^^ from sample to sample are probably due to
the number of variables involved in the sample preparation; 1) particle
size and degree of agglomeration, 2) sol concentration and wettability
of the sol, and 3) the amount of surface oxide on the starting SiC.
Infiltrations of SiC precursor polymers in the porous composite bodies
result in a little increase in Of^Q^ (Table IV-2) and Kjq (Table IV-7).
The density values suggest that very little silane is infiltrated under
the conditions used.
A linear type of relationship between a^-ig^ and p is shown in Fig.
IV-12. This means that one may need only density measurements to pre-
dict flexural strengths. Extrapolations based on the Fig. IV-12,
assuming a theoretical density of the composites of 60 vol% SiC:2.4 g/cc
for NC and 2.8 g/cc for SC, result in a theoretical a^^g^ °^ ^^^ ^^^^ ^°^
NC and 425 MPa for SC. These maximum theoretical values are still lower
than the experimental Of-^^^ of ~700 MPa obtained by Prewo and Brennan^^
215
by hot pressing borosilicate glass and continuous unidirectional Nicalon
fibers. This shows the difference between random fiber orientation and
uniaxial fiber orientation in af]g^. In the uniaxial fiber composites,
the fiber length to diameter ratio, R = «. The fact that a SC gives
much greater CTf]gx f^^^t be attributed to the R values, R ~ 3 for NC and
R ~ 10 for SC. This supports the theory^os that the reinforcement
effect of a composite is a function of the "aspect ratio," R. This is
consistent with results of Sambell et al.sa.iu for carbon fiber/glass
matrices composites.
No attempts, other than incorporating unidirectional continuous
fiber, were made to orient the chopped fiber in the Si02 matrix. The
resulting composites are thus expected to have randomly oriented fibers
and are treated as being isotropic. The volume fraction of SiC was kept
at a constant level of ~60 v/o for cast composites and ~70 v/o for
pressed composites. Variations in SiC loading were limited by the inher-
ent porous matrix of gel derived silica and monol ithicity of the compos-
ite body. The former limits the SiC volume loading toward a lower level
and the latter tends toward higher volume loading. This forces the SiC
loading to be at approximately a constant level (~60 v/o) of the satura-
tion to maintain monol ithicity.
Based on a small number of specimens used for a testing of thermal
shock resistance of the composites, both NC and SC have good thermal
shock resistance between 800°C and room temperature.
Figure IV-9 shows that heating the composites at atmospheric press-
ure up to 1400°C in a reducing atmosphere did not yield the theoretical
216
density. This imposes a major problem in densifying the composites.
Heating the composites in water vapor may nasten the densif ication pro-
cess but oxidation of SiC by the water may be a problem. A vacuum heat-
ing to 1400°C was shown to be detrimental to these composites by disin-
tegrating the composites. This is thought to be a result of chemical
reaction between Si02 and SiC according to Lq. IV-?.^"^' '''^
2 SiU^lS) + SiC(S) t 3 SiO(g) + CO(g) aG = -70 Kcal/mole (IV-7)
This is probably the main cause of the "cracker effect" observed for the
organosi Iane/Si02 gel composites in Chapter III as well.
Silicon carbide ceramics have modest fracture toughness, expressed
by the critical stress-intensity factor, Kjq : ~3 MPa •m^/^. ^ ^5. 116 jhg
NC and SC, however, have Kjq values as high as -8 MPa.m^'^ for SC and ~6
MPa«m^'^ for NC (Table IV-6). These are remarkably high toughness
values considering the modest Jfig^ of these composites. This high
toughness must be attributed to the action of yielding fibers which
bridged the crack surface. This is evidenced in the load vs. displace-
ment curves in Figs. IV-14.
Among the various mechanical properties, fracture toughness as
expressed by Kjq is considered as important a parameter in design con-
sideration as the various strengths. This parameter provides a measure
of tolerance of the material to presence of discontinuities. In some
applications, the fracture toughness is even more important than
strength. The fracture toughness relates to fracture strength, impact
resistance, and thermal shock and fatigue behavior. ii^ This implies
that NC and SC are potential candidates for heat-engine materials.
217
The generally lower mechanical properties of cold pressed NC than
tnose of cold pressed SC are attributed to the lower lenyth to dianeter
of fiber ratio (R).ii2 For Micalon^ R ~ 3 and R ~ 10 for Silar".
Figure IV-14 shows increases of fracture energy represented by the
area under the curves by silane impregnation into SC and into uniaxially
laminated NC. The initial sharp increase of load must be attributed to
the elastic behavior of the composites. Until the crack is propagated
to the fiber, no load increase is shown. This is the stage of the weak
matrix cracking. Then the sharp increase in load is the stage of fiber
controlled crack propagation and finally to failure.
Composites fired at 1400°C show a typical load vs. crosshead dis-
placement of a brittle material. The toughness curves shown in Fig.
IV-16 are essentially the same as those for flexural strength (Fig.
IV-15).
One of factor listed as important in mechanical properties of fiber
reinforced ceramic matrix composites is the interfacial bonding between
the fiber and the matrix. ^^'^ As a result of chemical reactions between
the fiber and the matrix, the bond strength between the two phases can
be increased and this can cause fiber degradation. Rice et al.^O'' have
shown that this fiber-matrix interaction can be controlled by a third
chemical reagent as an additive to the matrix or as fiber coatings.
Characterization of fibers, matrix, and fiber-matrix interactions thus
requires microanalytical techniques because the scale over the critical
processes occur is quite small; 2-3 ym across the interface.
Figure IV-17 shows NC after 80°C drying. The 59,000X TEM micro-
graph reveals that the particle-like gel matrix has a distinct interface
218
with Nicalon*^ fiber. It shows no matrix-fiber interaction at 80°C. The
STEM micrograph at 300X, however, shows debondings or cracks in the
interfaces. These cracks must have developed during the thinning pro-
cesses. This means that the bond between fiber and matrix is relatively
weak in cast SiC/Si02 composites. The composite after heating to 90U°C
in nitrogen shows the gel matrix is no longer particle-like as shown in
Fig. IV-18 at 50KX. The interface is certainly less distinct than the
composite at 80°C. This may be an indication of tne fiber matrix inter-
action. At IKX, cracks in the matrix are still present.
Figure IV-19 shows a SC sample after 80°C drying. The TEM micro-
graphs at 50KX shows a random orientation of Silar" whiskers bonded by
SiOp gel matrix with a distinct interfacial boundary. This is also
shown by the micrograph at 200KX. After 900°C heating, the extension of
the matrix into the whisker is not clear (Fig. IV-20). The whiskers
show a distinct columnar subgrain structure representing crytal 1 ographic
mi sorientations. This kind of planar stacking disorder is common in
SiC, partly because of the very low stacking fault energy {OUUl} of a
planes. ^13 Corresponding selected area diffraction patterns exhibit
spotty rings for the a-SiC single crystal.
The as cast surface of NC after a 110U°C heat treatment. Fig.
IV-21, shows the random orientation and distribution of the fibers in
the Si02 matrix. Figure IV-22 shows that the fibers and Silar" whiskers
are well bonded by the gel matrix after drying.
Scanning electron micrographs of fractured surfaces of NC with uni-
directional and random chopped fibers show fiber-pull-out as indicated
219
by the rough surfaces. This either disagrees with the work of Bender et
al.^^3 or an extensive matrix-fiber interaction may not occur at temper-
atures below ~1200°C. This may be another advantage of forming a SiC/
Si02 composite via sol-gel process. In the work of Bender et al.,ii9
they used hot-pressing to form Nicalon®/Si02 composite at temperatures
well above 1300°C. At this temperature, the fiber-matrix interaction is
expected to be quite extensive resulting in a significant reduction of
mechanical properties. 1^9
Rather large scattering of data points are seen in Table IV-1,
Figs. IV-9, IV-IU, IV-11, and IV-12. This is caused by the difficulty
of using one identical sample throughout the thermal treatments as well
as the characterization steps. Even each composite sample prepared
under identical experimental conditions tended to be different after an
identical thermal history. This is because there are other variables
present such as nonuniform property and distribution of the starting
materials, uneven distribution of flaws, impurities, etc. which affect
the properties.
As given in Table IV-1 and shown in Fig. IV-37, the mean Of^Q^
after treatment at 900°C in N2 for four hours is 92 ± 26 MPa for SC and
31 ± 6 MPa for NC. Under same condition but in air, af^g^ ^^ ^^ * 23
MPa for SC and 27 ± 5 MPa for NC. Both NC and SC exhibited -14% and
~13% reduction respectively in the strength by the oxidative exposure.
However, within the uncertainty range, this reduction in strength may
not be significant. Clark et dl.120 showed a reduction of Nicalon®
tensile strength from ~1900 MPa to ~8b0 MPa after treatment at 1UUL)°C in
220
110
100
90
- SC
in
^2
-
\\\^
■
\^
SC
in
Air
80
W^
\\^
^flex,
60
50
-
i
1
\V
^0
-
^
w
NC
in N2
NC in Air
30
R
20 .
WM
__
licalon/SiOp Gel and Silar/Si02 Gel Composites
Fig.IV-37. Flexural Strengths of SC and NC After Exposed
to Air or Nitrogen at 900 °C for 5 hours.
221
wet air for twelve hours. This ~80% reduction in strength may not De
used in direct comparison with HC because of the differences in the ex-
perimental conditions and the nature of the sample. NC was heated in
dry air at 900°C for four hours while the Nicalon fiber of Clark et
al.i20 \^as heated at 1000°C for twelve hours. Moreover, the testing
method is tensile vs. bending. However, it may be inferred that Nicalon
fibers in NC are more resistant to oxidation than the naked Nicalon
fiber in a similar environment.
Figure IV-38 compares flexural strengths of NC and SC with some
other ceramic composites of comparable composition fabricated by other
investigators as given by the reference numbers. Comparisons here are
based on the most recent literature and not necessarily on the highest
values available in all the literature. The purpose of these compari-
sons is not to show the superiority of the NC and SC of this work over
other composites of similar composition but to show tnat NC and SC are
not inferior or, rather, tnat they are comparable in some cases. These
other ceramic composites were mostly fabricated by hot pressing techni-
que. A direct comparison shows that SC and NC are generally inferior to
other composites but superior to tnose of Gac et al.''^ and Wilson and
Breit.121
The lower values for NC and SC are due to lower densities. When
this is taken into account by calculating the a^ig^/p ratio, the values
for NC and SC become comparable with other ceramic composites as shown
in Fig. IV-39. Figure IV-39 compares the ratio of flexural strength to
density of NC and SC with various other reinforced ceramic composites by
222
600
500
400
(T ,
flejj
MPa
300
200
SiC fiber/
Vycor
THIS WORK
Cold
Pressed
Cast MC
(92)
(79)
30v/o SiC/
tnullite
(78)
SiC/boro-
(121) silicate
SiC whisker/ glass
Si02
^^
100
CERAMIC COMPOSITES
Fig.IV-38. Flexural Strengths of NC and SC as Compared with Some Other
Ceramic Matrix Composites
223
c
3
O
u
Qi
M
CO
•^
J=
?
4)
4J
/^ D
1—
0^ ^
^m
r«. to
^^
N.^
^
S
>
o
PI
u
Q)
X
T-t
-'-^
U- M
fo
W
CN
ncc
I— i
C .-'
'--^
CNO£
V)
u
4->
•f—
Ul
O
Q.
E
O
<_3
o
>pp.
E
ro
s-
c;
(J
s-
a>
-C
1-j
O
CO
^
IzJ.
+->
•^
:/■)
<u
^
c ■-
rtj
c;
a.
t=
■><
o
— ^
(_j
cc
(-
u)
<
(T3
7-
<_)
U
OO
K^
T3
^
C
^
fO
u
o
o
m
CM
O
c
csj
Tt-
U
X
0)
ac
o
o
o
in
c
o
c
Ol
224
other investigators as given by the reference numbers. Cast or cold-
pressed NC and SC of this work are comparable with other composites of
similar composition fabricated by hot pressing technique. The Nicalon'^/
Vycor® composites of Prewo and Brennan92 show exceptionally high a^-, /
p. This exceptional property is, in part, resulted from the uniaxial
orientation of continuous Nicalon fiber and two-step process i.e. pre-
form the composite using wet traditional ceramic processing then hot
pressing. Lange et al.82 showed 1340 MPa in Of-^^^ value for their ZrO^/
AI2O3 in a four step process. On the other hand, NC and SC were fabri-
cated in a one step process.
Figure IV-40 compares Kj^ values of iNC and SC (Table IV-6) with
other comparable composites. Fracture toughness of values of NC and SC
compare well with those of other composites. Tne Kjq of NC and SC are
generally superior to the other composites with exception of those of
Prewo and Brennan9i and Ricei23 for SiC fiber/glass matrix composites.
The SiC fibers in these exceptional composites were unidi rectional ly
preformed followed by hot pressing procedure while NC and SC are random-
ly oriented short fibers and whiskers without hot pressing. Even more
exciting comparison of NC and SC with other ceramic composites for the
ratio of Kj^ to density is shown in Fig. IV-41.
It has now been shown that mechanical properties of NC and SC are
comparable or superior to many other "the state-of-the-art" ceramic com-
posites of similar composition made by more conventional, tedious, and
restricted techniques. Moreover, fabrication of SiC using sol-gel tech-
nique still has a room for improvements as discussed previously.
The composites NC and SC upon heating at 1100°C in air caused the
oxidation of SiC as shown in Figs. IV-24 through IV-28. At 11U0°C, both
225
1/1
o
O
«_1
CD
01
s.
Q.
E
O
<J
I/)
CT3
O
■a
c
t/i
3
o
Ol
S.
3
4->
U
S-
o
I
226
o
•a
s_
C
E
O
on
c
c: o
o
on
c l/l
1/1
■t-' o
Q.
lO E
i/l o
01 (_}
c
^ X
3 S-
O -M
I— TJ
OI
s- o
3 •■-
4-> E
O (O
<a s_
OI
227
the vitreous SiOo cind Cristobal ite can be formed as products of SiC
oxiddtion. Cristobal ite formation rapidly progressed as indicated by a
reflection band at -1170 cm"^ (R^) for both s-SiC (Nicalon) and a-SiC
(Silar). Figures IV-29 through IV-32 illustrate this phenomenon more
vividly.
A large amount of vitreous SiOo matrix in the composite should sup-
1100°C
press the forward reaction of the oxidation, a-SiC — t^ *■ vitreous
^2
SiOj. As a result, formation of cristobalite over vitreous Si02
predominates. This was supported by an XRD analysis of the heat treated
composites.
For Nicalon® (B-SiC) in silica matrix, the rate of cristobalite
formation, which is a direct result of oxidation of SiC at 1100°C,
rapidly increased up to ~5 hours then decreased up to ~10 hours (Fig.
IV-29). The rate increased again after ~10 hours and slowed down after
~15 hours. On the other hand, the raw Nicalon (no matrix) showed a
continuous increase of the rate of cristobalite formation.
A similar but less dramatic behavior is shown for the Silar"
(a-SiC) composite. Silica crystal formation rate for the raw Silar"
continued to increase, however (Figs. IV-31 and IV-32).
These complex oxidation rates of SiC may be explained by works of
Pampuch and coworkers. 12"+. 125 They showed a growth of a ternary inter-
mediate species Si-O-C which could not be etched by HF.
In Nical on/Si O2 composite, a rapid increase of R^/'^q ^^^^ ^^"^^ ^P
to ~5 hours may be an indicative of the initial oxidation of SiC. As
the Si02 film from the initial oxidation of SiC passivates or decrease
228
the oxidation rate, the reaction of SiC + SiUg to yield SiOC may take
place after ~5 hours of an oxidative heat treatment of Nicalon/SiO^
composite. In that case, the amount of Si02 and SiC would decrease as
long as the reaction rate to give SiOC is appreciable. However after
~10 hours of oxidative exposure, the rate of Si02 growth increases and
slows down after ~15 hours. This may be the period of a reaction
SiOC +02^ Si02 + CO, (IV-8)
thus increasing Ry/Rr, and decreasing R./R^. The increase in R^/R^ after
Ay i_ y ■- y
~15 hours, however, may be due to a development of microcracks in the
matrix by evolution of gaseous products, e.g. CO, CO2. This is also a
cause of the continuous oxidation of SiC after ~2U hours, although at a
slower rate.
The TGA data of SiC and the composites as given in Table IV-8
indicate that the weight gain due to SiOo formation per unit weight of
sample is much smaller for SiC in a SiOo glass matrix than the raw SiC.
In 5 hours, raw Nicalon® (s-SiC) gained more than twice the % wt than
did the Nicalon fiber in a glass matrix. The Silar" (a-SiC) whiskers in
a glass matrix are nearly BOX more resistant to oxidation than raw
Silar" within 12 hours of oxidative heat treatment.
The ^J^n vs. time plot for Silar" given in Fig. IV-32 does not
show the downward curve as does Nicalon®. However the Ry/Rr, values due
X y
to oxidation of Silar" somewhat flattened between 10-15 hours of oxida-
tive exposure. This may mean that the formation of the SiOC intermedi-
ate phase in Silar™ is not as extensive as in Nicalon. The period of
SiOC also came somewhat later (~10 hours) in Silar" than in Nicalon® (~5
229
hours). It appears that a little longer time is needed for Silar" to be
passivated than for Nicalon. This may be due to the greater SiOg con-
tent in as-received Nicalon® than in as-received Silar™.
Therefore, the ternary intermediate SiOC phase is formed between
~5-10 hours for Nicalon® and ~10-15 hours for Silar" in porous Si02
glass matrix. However this SiOC is formation was not detected for raw
SiC samples (Figs. IV-30 and IV-31). This can be explained by the phys-
ical properties of the SiC materials. Silar" is a single crystal a-SiC
with greater density (3.2 g/cm-^) of hexagonal crystal and greater pur-
ity, while Nicalon® is a microcrystal 1 ine 6-SiC which has a cubic form
with a density of 2.6 g/cm^^ and substantial amount of graphitic carbon,
Si02, and Si as impurities. These impurities should act as oxygen scav-
engers hindering oxygen diffusion to reach SiC unlike al uminosil icates,
where the aluminosil icate impurity increased the oxidation rate of hot-
pressed SiC by enhancing oxygen diffusion. 126
It appears that the ternary intermediate SiOC exists between SiC
and Si02 interface^os and unless the SiOo blocks oxygen diffusion
completely, the reaction of SiOC + O2 t Si02 + CO should continue until
the Si02 filni brings complete passivation. It seems more likely then
that oxygen diffusion is the rate limiting step rather than CO removal.
The trend of deceleration of the oxidation rate of SiC in the gel
derived glass matrix as shown in Figs. IV-29, IV-31, and IV-32 after ~5-
10 hours has been explained by the formation of a SiOC ternary inter-
mediate phase. This similar trend was also observed by Suzuki et al.i27
in their study of a-SiC oxidation. They observed a break in the curve
230
of increasing SiO^ layer thickness after ~6 hours of oxidative exposure,
but no explanation as to what caused the break was given. Leei23 also
has observed the similar trend of a break in the increasing oxidation of
SiC with time in oxygen atmosphere, but he treated this anomaly as an
artifact despite the repeated observations. Again a similar behavior of
SiC oxidation in a gel derived alumina matrix was observed by LaTorre et
al.^29 jhey 3]so gave no explanation for this behavior.
Conclusion
Low density SiC/SiOo composites can be prepared easily by a sol -gel
method without hot pressing. Complex shapes can be easily cast using
SiC particles dispersed in the SiO? sol.
The ease of forming and fabrication, stability at high temperature
with a moderate flexural strength, high fracture toughness and high
hardness have been demonstrated. It was further demonstrated that NC
and SC have attractive strength-to-weight ratios and even more attrac-
tive toughness-to-weight ratios, good oxidation resistance, and good
thermal shock resistance, yielding a potential ability to withstand
severe environment. Achieving the full density of the composites will
undoubtedly improve the properties. Considering the low density of the
composites tested (p ~ 1.9 g/cc), the toughness observed (Kjq ~ 6-8
MPa«m-'-'^) is truly remarkable. The Kjp of NC and SC is superior to many
other ceramic composites and the Kj^/p values of NC and SC are superior
to all but highly oriented hot-pressed samples.
The high porosity of NC and SC is due to the large inherent shrink-
age in the silica gel matrix. The flexural strength was shown to be
231
directly related to the density of the composite body. The lower den-
sity and lower strengths of the Silar" system indicate that the Si02 sol
is less compatible with a-SiC Silar" than with B-SiC Nicalon®. This
must be because of the amorphous nature of Nicalon and the large amount
of Si02 and graphite present in Nicalon®.
Although an improvement in density and strength can be achieved by
multiple impregnation of Si02 sol into the micropores of the composite
bodies by dipping, there is a limit on the improvement since the impreg-
nation is confined to a the surface region after ~4 cycles.
Impregnation of the porous NC and SC with polysilanes, SiC precur-
sors, followed by a pyrolysis show a little improvement in the flexural
strength after each cycle.
Nicalon® and Silar" SiC have less oxidation in air at 1100°C when
SiC is incorporated in a sol-gel derived Si02 glass matrix than when raw
SiC is exposed to sane condition. The oxidation kinetics of SiC in Si02
matrix (p = 1.8 g/ml ) after heating in air at 1100°C is 3-5 times slower
than the pure SiC under same conditions. However due to the residual
porosity, complete immunity of SiC oxidation in the matrix in the begin-
ning was not observed.
Both Nicalon® and Silar" in a Si02 porous glass matrix may be oxi-
dized via formation of a ternary intermediate SiOC phase. The oxidation
reaction mechanism of SiC in porous Si02 glass matrix appears to be com-
plex and requires further work to understand completely. Silar" (a-SiC)
is more prone to oxidation than less pure Nicalon (g-SiC). This may be
due to the larger surface area of Silar" than Nicalon®.
232
In order to improve the oxidation resistance of SiC further, it is
necessary to reduce the porosity of the composite and prevent natrix
damage after a prolonged oxidative exposure at high temperatures.
Some immediate applications of this material may be in fusion power
reactors. Silicon carbide material because of its low plasma contamina-
tion, low induced radioactivity, capability of high operation tempera-
ture, and relatively abundant raw material supply, may be a leadiny can-
didate in the plasma chamber of a fusion machine. A formed SiC/Si02
bulk composite may be used directly or as a metallic part coated with
this composite. As mentioned in this chapter, coating of a material is
expected to be as simple and easy as casting.
Another area of application may be in high temperature radomes.
Requirements for high temperature radomes demand a material with a high
toughness, good thermal shock resistance, high resistance to rain and
sand erosion, and light weight. The composites described in this
chapter should meet these requirements.
The high values of Kjq/p and/or Of^^^/p of these materials may be
especially significant in space applications such as materials for ad-
vanced spacecrafts, space transport systems, and large scale antenna
arrays, which require high thermal performance and light weight. The NC
and SC should provide significant improvement in thermal stability and
mechanical properties and be capable of in-space processing. It should
be possible to control the elastic modulus and stiffness of such space
structures and their damping capacity by varying the volume fraction of
fibers or whiskers in the composites.
CHAPTER V
OTHER CHEMICALLY DERIVED CERAMIC COMPOSITES
Introduction
The advantages of chemically derived ceramics and composites have
been presented in the previous chapters. In this chapter, processing of
several other chemically derived ceramic composites of mullite fibers
and gel derived silica, alumina powder and polysil astyrene, SiC/SiC com-
posites of Nicalon® and Silar" with polysi 1 astyrene, and molecular com-
posite powders of SiC/TiC and SiC/Al203 are presented.
The mullite fiber is commercially readily available as Nextel™*
which is derived from a sol-gel process. A monolithic composite incor-
porating Nextel" into a sol-gel derived silica matrix is attempted simi-
lar to the processing of the SiC/Si02 composite in decribed Chapter IV.
Using the processing techniques of SiC from PSS as described in
Chapter II, it is anticipated that a SiC fiber/SiC matrix composite can
be made. A successful fabrication of SiC/SiC composite should yield
many desired properties.
Wei and Secher^^s ^ere able to increase Of^Q^^ from 500 MPa to 680
MPa and Kjq from 4 to 6 MPa.m^/^ by incorporating fine TiC particles in-
to a SiC matrix followed by hot pressing at 2000°C. Yajima et al.i2
have obtained SiC/TiC powder by mixing their polycarbosil ana with a
Ti(0R)4 followed by pyrolysis to above 1500°C. Hence, it is hoped to
synthesize a SiC/TiC molecular composite powder using polysilastyrene to
4= Manufactured and provided by 3M Company, St. Paul, MN.
233
234
demonstrate that the desired ceramic naterial can be obtained by
chemical processing.
Silicon carbide/AlgOj composites have been made by Wei and Becher-'i
and by Cutler et al.^^ by hot pressing mixed powder of SiC and A1203. In
this chapter, a novel way of synthesizing SiC/Al203 composite powder at
the molecular level is demonstrated by mixing the precursors of SiC and
AI2O3.
Experimental
Nextel" 312 continuous fiber which has a mullite composition was
provided by 3M Co. Nextel" fibers were cut and stacked unidi rectional ly
with a desired thickness in a rectangular plastic mold. The fiber stack
was filled with silica sol as described in Chapter 4. After covering
and sealing the mold, it was aged, dried, and infiltrated with the sol
4X as for Nicalon® composites in Chapter 4. After an appropriate heat
treatment in air, density and flexural strengths were measured the same
way as described for Nicalon® and Silar" composites in Chapter 4.
Alumina/silicon carbide composites were prepared by mixing 3 g of
Baikalox CR6* fine powder with 5 ml of 20 w/v PSS solution in benzene
followed by cold pressing in the steel die as shown in Fig. IV-4. The
cold pressed green bodies were pyrolyzed to convert PSS to SiC at 80U°C.
Silicon carbide/silicon carbide composites were prepared by mixing
3 g of chopped Nicalon® or 3 g of Silar" with 5 ml of 20 w/v PSS solu-
tion in benzene and cold pressing as above.
* a-Al203 with 1 ym particle size by Baikowski International.
^35
A molecular composite of SiC/TiC powder was prepared by mixing 10 g
J-PSS2, 0.82 g DCP, and 8.5 ml of titaniun isopropoxide in 60 ml of tol-
uene under ^2 at 110°C in an apparatus shown in Fig. II-3 and a gentle
reflux condition for 14 hours. The solution had a dark green color at
the end of 14 hours. The product was precipitated out by ethanol and
washed with ethanol three times before vacuum oven drying. The dried
and agglomerated product had a bright orange color. Nearly 100% yield
was achieved.
A molecular composite of SiC/Al203 was prepared in a similar way to
SiC/TiC above but 0.75 9 DCP and 9 ml of aluminum sec-butoxide with
refluxing for 22 hours under No. The product was precipitated out by
ethanol and washed twice with methanol followed by vacuum oven drying at
150°C. The dried agglomerate appeared dull white in color and ~70%
yield was achieved.
Fourier-transform IR spectra before and after pyrolysis of the com-
posites were taken oy a ilicolet MX-1. Hot stage XRD by a Philips Elec-
tronic Automated X-ray Powder Di ff Tactometer with HTKIO Hot Stage was
used to obtain x-ray diffraction patterns of the molecular composites.
The composite powders were introduced on a platinum substrate using a
polymeric glue. Resistive heating was applied to the substrate in helium
atmosphere.
Results
Mechanical properties of Nextel" composites and Al203/SiC compos-
ites are given in Table V-1. Char yields of molecular composites pow-
ders are given in Table V-2. Energy dispersive spectra of PSS/A1(0R)3
236
Table V-1. Mechanical Properties and Densities of Nextel'"/Si02
Composite, SiC/SiC Composites, and Alumina/SiC Composites.
Composites-t
afiex, MPa
Nextel (60)/
SiOg, 900°C
29±3
Silar (70)/
PSS, 800°C
15±2
Nicalon (70)/
PSS, 800°C
30+3
AloOo (75)/
PSS, 900°C
22±2
KjQ, MPa m^/^
0.43±1
2.41±1
DPN
p, g/cc
1.92±0.1
434±28 1.60±0.1
55U32 1.65±0.1
1.87±0.1
* Three specimens used for each composite. The numbers in parentheses
are voli in the pyrolyzed composite.
237
Table V-2. Char Yields of Molecular Composite Powders
Composite
Wt%
SiC
% Weight Char Yield
at 940°C in N2 for 14 Hours
PSS/Ti(0R)4
PSS/A1(0R)3
30
35
25±6
55±7
238
and PSS/Ti(0R)4 before pyrolysis are shown in Fiys. V-1 and V-2. They
show the constituent elements Si; Al and Si; Ti respectively.
Fourier-transform IR spectra of PSS/A1(0R)3 after drying and after
pyrolysis at 900°C are shown in Figs. V-3 and V-4. Figure V-5 shows
FT-IR spectra of PSS/Ti(0R)4 composite before and after pyrolyzing at
940°C.
Scanning electron micrographs of PSS/A1(0R)3 and PSS/Ti(0R)4 after
pyrolyzing at 940°C in N2 and hand-grinding are shown in Fig. V-6. They
all show agglomerates.
Hot-stage XRD of the pyrolyzed molecular composite powders are
shown in Figs. V-7 through V-15.
Di scussion
Nextel" fiber is not a strong fiber. The tensile strength data,
provided by the manufacturer, 3M Co. is ~10 MPa. The ofigx °^ Nextel/
Si02 composites after multiple sol impregnation and 90U°C sintering in
Table V-1 is modest after considering the relatively low tensile
strength of the fiber. The significance here is that Nextel" fibers can
be formed into monolithic structural composites if necessary. A higher
temperature heat treatment should improve density and, thus, mechanical
properties.
The SiC/SiC composites made from Silar"/PSS and Nicalon®/PSS after
pyrolyzing at 800°C yield a somewhat low af-jg^ mainly because of the
high porosity. Sintering of these composites to full density is expect-
ed to be difficult by the reasons given in Chapter 2.
239
o
m
CD
CD
>
UJ
O^
l/i
= o
o
Q.
ihc^J
E
o
6
(_3
CO
a;
o
V — ■
r—
<
v~>
oo
ID
a.
ID
M-
in
O
oJ
E
3
S_
-(->
u
QJ
Q.
OO
oo
Q
LU
o
o
1—
o
>
d
CTl
240
<u
o
a.
E
o
o
to
O.
Z3
S_
■*->
o
<u
Q.
to
to
a
CM
I
241
ro
GO
in
00
o
o
o
CVJ
029
S6> 652
30NV103nd3a 7o
t722
680
lO
(M
(U
o
a.
o
o
in —
^— • 1
E
Di
o
O
^'— '
1—
CO
<
(T
I/O
ro LU
n
ro CD
12
3
tsl
z
to
UJ
>
4->
<
C
h-^
>>
1
f^
to
<
o
E
=s
&-
+->
u
O)
t/0
CO
I
242
98l£
30NV103nj3a 7o
295"
CD
O
S-
o
o
CTi
(.1
>i
o
!w
>>
D.
S_
Ol
+->
o
Q.
14-
O
u
Q.
1201
in
I
243
4400
3200
2000
1400
800
WAVENUMBERS (cm"')
Fig.V-5. FT-IR Spectra of PSS/Ti (OR) After Drying at 60'C and Pyrolyzing
at 940'C ^
244
Fig.V-6. SEM Micrographs of PSS/AU0R)3 (top) and PSS/Ti(0R)4 (bottom)
After Pyrolyzing at 940°C
245
Pt
1
Pt
..-.Jj.
1
1
/^V-^jW^X/^Vv-yy-J V^
1
1
20
26
32
1 1 —
38
Pt
1 1 1 —
44
1 U
50
50
56
68
74
80
«
c
«
29
Fig.V-7. Hot Stage XRD Pattern SiC/Al203 Composite, Top: at RT After
Pyrolysis at 940''C in N2, Bottom: at ISSO^C after 10 min.
246
Pt
Pt
20
26
l^A\,,A<^AVl«A^yvW'
32
38
Pt
50
-I 1
50
6S
SO
(fl
c
20
26
32
Pt
3S
Pt
Pt
50
50
56
68
29
74
80
Fig.V-8. Hot-Stage XRD Pattern of SiC/Al203 Composite,
Top: at ISSO^C after 50 min.
Bottom: at 1400°C after 12 min.
247
(0
C
0)
26
Pt
48
56
Pt
38
Ttr
Pt
Pt
64
72
IB
50
80
Fig.V-9. Hot-Stage XRD Pattern of SiC/Al203 Composite, Top: at 1400 C
After 30 min.. Bottom: at RT After 45 min. at 1400 C
248
iq V-10. Hot-Stage XRD Pattern of PSS/Ti(0R)4 Composite, Top: at RT
After Pyrolysis at 940°C, Bottom: at 900°C After 10 min.
249
20
7'
c
^^6^11
WjVV'^'VvA'
26
32
38
Pt
56
62
68
4 A
74
50
L
80
Fig.V-n. Hot-Stage XRD Pattern of PSS/Ti(OR), Composite, Top: At 900"c
After 40 min,, Bottom: At IIOO'C After 10 min.
250
Fig.V-12. Hot-Stage XRD Pattern of PSS/Ti(0R)4 Composite, Top: at 12G0*C
After 13 min.. Bottom: at IZOCC after 30 min.
25'
20
26
32
Pt
44
50
^0
56
62
68
74
80
(0
C
0)
Pt
20
26
32
38
44
50
Pt A
50
56
62
68
74
80
20
ng.V-13. Hot-Stage XRD Pattern of PSS/Ti(0R)4 Composite, Top: at 1300'C
After 12 min.. Bottom: at 1300*C After 30 min.
252
20
^
(0
c
TieOii
26
56
PC
32
38
Pt
62
68
74
50
80
Fig.V-14. Hot-Stage XRD Pattern of PSS/Ti(OR), Composite, Top: at 1360*C
After IC min.. Bottom: at 135C°C After 30 min.
253
Pt
20
>
26
32
Pt
50
(050
Z
Ul
I-
z
56
62
68
74
80
TigO^^ or
SiC or Tie
Pt
Pt
16
Tt
Pt
32
40
■^
40
48
20
64
72
Fig.V-15. Hot-Stage XRD Pattern of PSS/Ti(0R)4 Composite, Top: at 1350
After 1 hr.. Bottom: at RT After ISSO^C for 1 hr.
254
Alumina/SiC composites formed from a-Al203 fine powder and PSS fol-
lowing pyrolysis at 80U°C show similar mechanical properties and densi-
ties to SiC/SiC composites above. Since SiC phase is the matrix, the
sintering may be difficult. More work is needed to show the strength as
a function of heat treatment temperature.
The char yields of PSS/Ti(0R)4 and PSS/A1(0R)3 composites are mod-
est to low. These are crosslinked polymers which are insoluble in com-
mon solvents. Yajima et al.^^ postulated that their polycarbosil ane is
crosslinked via Si-O-Ti. In the PSS system, because of the absence of
hydrogen directly bonded to Si, the crossl inking was expected to occur
via Si-CH2-0-Ti and via Si-CH2-0-Al for PSS/A1(0R)3.
These are only speculations and there is no way to know the precise
nature of the chemical reaction between PSS and A1(0R)3 and between PSS
and Ti(0R)4 without further studies. Further study varying the reactant
ratio and amount of the CFRI should be helpful.
Energy dispersive spectra in Figs. V-l and V-2 show the constituent
metallic elements in the green polymers.
Fourier-transform spectra of PSS/A1(0R)3 in Figs. V-3 and V-16
resemble the spectrum of PSS more than that of A1(0R)3. However, the
Al-O-Al vibrational mode is shown at -1100 cm"^. The band at 1100 cm"
became more prominent after the composite was pyrolyzed to 900°C (Fig.
\/-4). A reflectance response representing the Si-C stretching mode is
shown at -800 cm"^ as a step. In Fig. V-4, it can be seen that the
pyrolyzed PSS/A1(0R)3 composite indeed has Al-O-Al and Si-C as expected.
Figure V-5 shows the FT-IR spectra for PSS/Ti(0R)4. Most of Si-CH3
features in PSS are lost after crossl inking with TilOR)^. After
255
Al (Sec-Butox)3
o
z
o
UJ
on
J- pss/AKSec-Butox),
5600
4400
800
200
3200 2000 1400
WAVENUMBERS, cm"'
Fig.V-15. FT-IR Spectra of PSS, A1(0R)3, and PSS/A1(0R)3 After Drying
255
pyrolyzing at 940°C, the bands for Ti02 (-750 cm"^) and SiC (-800 cm"^)
are shown. The conversion of Ti02 phase to TiC phase is only possible
at temperatures above HOO^C. Yajima et al.^^ showed the transformation
of Ti02 to TiC at 1700°C. Thus at 940°C, the composite must consist of
SiC and Ti02 and that is what is shown in Fig. V-5. This is documented
by hot-stage XRD as shown in Fig. V-10. A small amount of titanium
dioxide, actually TigO^^^^, is formed below at ~950°C. As temperature of
the composite in He is increased, the XRD intensity for TigOii did not
increase. The peak at 54.3° 2e (1.69 A) cannot be identified. It is
not SiC, TiC, Si02, AI2O2, nor Ti02. As temperature increased to 1100°C
(Fig. V-11), a new and sharp peak appeared at 52.5 29 (1.74 A). This
peak may be an artifact or an intermediate species because it dis-
appeared immediately after raising the temperature. At a temperature of
1200°C, the peaks for TigO^^ are diminished (Fig. V-12). A further
diminishing is observed at 1300°C and after 30 min at 1300°C, they all
disappeared. At 1360°C, even platinum peaks are diminished. After one
hour at 1360°C and cooling to room temperature, the small peak at 36° 29
representing TigOii came back. However the position of 36° 29 is the
position for SiC as well as TiC. The true identity of the peak at 36°
29 is not clear. The reason why the TigOii crystalline phase dis-
appeared at 1300°C is unknown since the m.p. of Ti02 is ~1840°C. This
may be the temperature at where the transformation from Ti02 to TiC and
solid state reactions among the components might begin.
It appears that the crystallization of TigO^]^ is somewhat
suppressed as discussed in Chapter 3 for Si02/SiC composites. This
257
phenomenon is even more strongly demonstrated in the PSS/A1(0R)2
composite. Figures V-7 through V-9 show that no crystalline phase is
present in the SiC/Al203 composite at a temperature as high as 1400°C
for 40 min except the peaks for the Pt substrate.
Figure V-6 shows that the SiC/Al203 and SiC/TiOg composites powders
are agglomerated.
Conclusion
Composites of Nextel" fiber/SiOg gel and Al 2O3 powder/SiC from PSS
can be formed with modest flexural strengths. Silicon carbide/silicon
carbide composites can also be formed easily and also have modest mech-
anical properties. Temperatures higher than ~900°C for sintering should
improve the density as well as the mechanical properties of these com-
posites.
The moleciilar composites of SiC/Al203 and SiC/Ti02 are formed from
chemicals of metal -organic derivatives. They appear to have a submicron
particle size and suppress crystallization of each phase in the compos-
ite at temperatures up to -1400°C. A higher temperature than 1400°C is
required to transform SiC/Ti02 to SiC/TiC.
The primary and common problem associated with the physical and
mechanical properties of the composites produced herein is the densifi-
cation. Use of sintering aids, a controlled atmosphere with higher
temperature capability should help the problem.
It has been further demonstrated in this chapter that the potential
to produce desired ceramic materials, which are difficult to obtain by
258
the conventional processing, via chemical process is enormous. However,
this enormous potential can only be made useful through a continueo
understanding of the chemical processes involved and elimination of
uncontrolled porosity.
The significance of this chapter should be in tne concepts rather
than a production of exciting mechanical properties of these composites.
The concept of mixing two precursors of a composite at a molecular level
can be applied to many other systems. Testings for mechanical proper-
ties of these composites must be followed after a successful synthesis
of a molecular composite is made.
CHAPTER VI
CONCLUSIONS AND RECOMMENDATIONS
Silicon carbide material can be made from silicon and carbon con-
taining polymers in a relatively easy manner. These precursors can be
used to strengthen the brittle materials such as glass by forming com-
posites. These composites show increased strength and fracture tough-
ness even with the low temperature heat treatments and low densities
(1.7-2.1 g/cc).
It becomes rather clear that there are variety of functional groups
which may be introduced to the polysilane backbone. A systematic study
of how these groups are related to char yield of silicon carbide, vis-
cosity, green density, chared density, susceptibility to oxidation in
green state and in the pyrolyzed state and crossl inkabi 1 ity is recom-
mended.
Changing the porosity and pore size of silica gel matrices could be
used to vary the amount and depth of SiC precursor silanes and thereby
improve the control over the physical and mechanical and/or optical
properties of silane impregnated silica glass monoliths. To achieve
this control, the behavior of pure silica gel monoliths with respect to
pore collapse and densi fication must be better understood.
Use of a nonpolar hiyh vapor pressure solvent should help to in-
crease the loading of the SiC phase in monolithic SiC/SiU2 molecular
bulk components. However, the increasing difficulty of maintaining
259
260
monol ithicity of the composites with increased loading of the SiC pre-
cursor needs to be investigated as well.
The chemical reaction between the SiC precursor and the Si02 gel
matrix is a serious problem and needs to be understood. A systematic
study of the reaction should help to produce better composites and is
highly recommended.
Although the Nicalon/Si02 and Silar/Si02 composites have a remark-
able strength-to-density ratio and toughness-to-density ratio, further
study to consolidate the bodies is needed. Complete consolidation may
be possible using a controlled atmosphere high temperature furnace.
The molecular composite powders of SiC/Al203, SiC/Si02, and SiC/
Ti02 should be hot pressed to determine their mechanical behavior. The
SiC/Ti02 composite may be converted to SiC/TiC composite by heating at
temperatures of ~1700°C. A TEM study to examine how the two phases are
arranged would be helpful in optimizing the process. This idea of mol-
ecular composite may also be applied to other systems such as Zr02 and
Tinted and tempered glasses may be made by an impregnation of the
SiC precursors within the ultraporous Si02 gel glass matrix. This glass
should produce very high strength and toughness with good IR absorption
characteristics which may conserve heat for example in cooking wares.
The high temperature limit of these wares may be raised to ~1500°C which
is 400-500°C higher than that of pure silica glass ware. Thermal shock
resistance of these composites due to the surface compressive stress may
be even higher than the pure silica glass.
261
Although the mechanical properties of these composites derived from
chemical processes described in this dissertation are far from the
theoretical limit, the possibility of closing the gap has been demon-
strated to be real through the continuous understanding of the chemistry
in the processes. The results are encouraging for the first attempts
and values for af^g^^/p and Kj^/p are equal to or surpass all but the
uniaxially oriented SiC fiber/glass hot-pressed composites.
As the science of materials has developed it has moved more from
giant leaps into the unknown to small steps forward in fairly clear di-
rections. The small steps are many times more sophisticated and repre-
sent a new level of scientific achievement.
It is hoped that the small understandings of ceramics and molecular
composites presented in this dissertation may accelerate the movements
toward clearly defined targets which we know will be achieved.
Some of more notable principles that were learned from this
research for Chemically Derived Ceramic Composites may be summarized
below.
1) A precursor to SiC, known as polysilastyrene (PSS), had been
shown to be potentially useful and superior to the existing precursor
(polycarbosi 1 ane) if it can be crossl inked without adding oxygen to the
polymer chain. First it was learned that PSS can be routinely synthe-
sized in a one-step process with a relatively straight forward manner.
2) Crossl inking of PSS can be achieved by using a chemical free
radical initiator. Through the crossl inking process, production of a
SiC monolithic body is possible without hot-pressing or high temperature
sintering.
262
3) Increased crossl inkabil ity of an organosilane is achieved
through reactive functional groups such as
4) These organosil anes can be used to impregnate porous ceramic
bodies such as silica gel followed by in situ crossl inking to yield a
SiC/Si02 composite. The subsequent pyrolysis leads to hardening effect
as well as a toughening and strengthening effect by the dispersed SiC in
the surface layer of the matrix.
5) The organosilane precursor can also be mixed with silica sol,
followed by a cogellation, by an in situ crossl inking, and by a pyroly-
sis to yield bulk SiC/SiiJ2 composites.
6) There is(are) a chemical reaction(s) between the impregnated
silane and the silica matrix which weaken the silica glass network.
These chemical reaction(s) must be identified and controlled for improv-
ed mechanical properties of silane/Si02 gel composites.
7) The sol -gel process to obtain SiO^ glass can be used to form
fibrous SiC/SiOo composites by dispersing the whiskers (or fibers) in
the silica sol followed by castings, gellating, or cold pressing, and
sintering. Characteristics of these composites are: a) low temperature
processed (~200-5G0°C lower than the conventional hot-pressing process)
so there is little damage on the fibers, and matrix and fiber inter-
action is minimal; b) low density materials (~1. 7-2.1 g/cc); c) mediocre
flexural strengths but good flexural strength to density ratio; d) sup-
erior fracture toughness represented by the critical intensity factor,
KjQ. Even more superior in Kj^ to density ratio to other ceramic com-
posites of comparable composition, e) This remarkable Kjq value comes
263
from the load transfer from the matrix to the fiber as well as tne mech-
anism of crack wandering and branching by pores. It was learned that
the fracture toughness is limited by the diameter to length ratio of the
fiber, f) Higher thermal shock resistance and better oxidation resist-
ance than the SiC not in the composite, g) Complex shapes can be formed
easily in plastic molds. The shapes and sizes are only limited by the
molds.
8) Molecular composite powders of Si02/SiC, Ti02(TiC)/SiC, and
AlnOo/SiC can be made by mixing the two precursors of the respective
phase with in situ crosslinking by a chemical free radical initiator.
These composites appear to have a high devitrification temperature as
well as high purity.
9) These technques: crosslinking, impregnation, sol-gel processing
for monolithic ceramics or glasses, and molecular level mixing of two or
more precursors can be applied to many other ceramic composite systems.
10) Silicon carbide incorporated into a porous silica glass matrix
is more oxidation resistant than raw SiC. The SiC in the composite
appears to be oxidized via an intermediate phase SiOC.
REFERENCES
1. Ultrastructure Processing of Ceramics, Glasses, and Composites, L.
L. Hench and D. R. Ulrich, eds., John Wiley and Sons, New York,
1984, pp. 15-557.
2. Better Ceramics Through Chenistry, C. J. Brinker, D. E. Clark, and
D. R. Ulrich, eds.. Materials Research Society Symposia Proceed-
ings, Vol. 32, North-Holland, New York, 1984, pp. 1-387.
3. C. D. Hendricks, Lawrence Livermore Laboratory Report, UCRL-8145,
Rev. 1, 1979.
4. K. J. Wynne and R. W. Rice, "Ceramics Via Polymer Pyrolysis" in
Ann. Rev. Mater. Sci., R. A. Huggins, ed., ]A_, 297 (1984).
5. B. E. Yoldas, J. Mater. Sci..^^, 1856 (1975).
6. B. E. Yoldas, Am. Ceram. Soc. Bull., 2, 2896 (1975).
7. S. R. Gurkovich and J. B. Blum, "Preparation of Monolithic Lead-
Titanate by a Sol-Gel Process," in Ultrastructure Processing of
Ceramics, Glasses, and Composites," L. L. Hench and D. R. Ulrich,
eds., John Wiley and Sons, New York, 1984, p. 152.
8. N. J. Arfsten, R. Kaufman, and H. Dislich, "Sol-Gel Derived Indium-
Tin-Oxide Coatings," in Ultrastructure Processing of Ceramics,
Glasses, and Composites," L. L. Hench and D. R. Ulrich, eds., John
Wiley and Sons, New York, 1984, p. 189.
9. E. Barringer, N. Jubb, B. Fegley, R. L. Pober, and H. K. Bowen,
"Processing Monosized Powders," in Ultrastructure Processing of
Ceramics, Glasses, and Composites," L. L. Hench and D. R. Ulrich,
eds., John Wiley and Sons, New York, 1984, p. 315.
10. K. S. Mazdiyasni, "Chemical Synthesis of Single and Mixed Phase
Oxide Ceramics," in Better Ceramics Through Chemistry, C. J.
Brinker, D. E. Clark, and D. R. Ulrich, eds.. Material Research
Society Symposia Proceedings, Vol. 32, North-Holland, New York,
1984, p. 175.
11. R. Rice, Am. Ceram. Soc. Bull,_62_, 889 (1983).
12. S. Yajima, T. Iwai , T. Yamamura, K. Okamura, and Y. Hasegawa, J.
Mat. Sci., 26., 1349 (1981).
13. R. Roy, presented at the Second International Conf. on Ultrastruc-
ture Processing of Ceramics, Glasses, and Composites, Palm Coast,
FL, Feb., 1985.
264
265
14. J. Persh, Ceramic Eng. and Sci. Proceedings , _5, 443 (1984).
15. "Silicon Carbide-1973," R. C. Marshall, J. Faust, and C. Ryan,
eds.. University of South Carolina Press, Columbia, South Carolina,
1974.
16. I. B. Cutler, "Production of Silicon Carbide from Rice Hulls," U.S.
Patent No. 3,754,676, assigned to the University of Utah, Salt Lake
City, Utah (Filed Oct. 30, 1970).
17. G. Jenkins and K. Kawamura, Polymer Carbons-Carbon Fibre, Glass and
Char, Cambridge University Press, London, 1976.
18. F. Ainger and J. Herbert, "The Preparation of Phosphorus-Nitrogen
Compounds as Non-Porous Solids," in Special Ceramics, P. Popper,
ed.. Academic Press, New York, 1960, pp. 168-181.
19. P. Chantrell and P. Popper, "Inorganic Polymers and Ceramics," in
Special Ceramics, P. Popper, ed.. Academic Press, New York, 1964,
pp. 87-103.
20. W. Verbeek, U.S. Patent No. 3,853,567 (1974).
21. G. Winter, W. Verbeek, and M. Mansmann, U.S. Patent No. 3,892,583
(1975).
22. S. Yajima, J. Hayashi , and M. Omori , Chem. Lett., 931 (1975).
23. S. Yajima, K. Okamura, and J. Hayashi, Chem. Lett., 1209 (1975).
24. S. Yajima, K. Okamura, J. Hayashi, and M. Omori, J. Am. Ceram.
Soc, _59_, 324 (1976).
25. S. Yajima, T. Shishido, and H. Kayano, Nature, 264 , 237 (1976).
26. S. Yajima, Philos. Trans. Res. Soc. London, A294, 419 (1980).
27. S. Yajima, J. Hayashi, M. Omori, and K. Okamura, Nature, 261 , 683
(1976).
28. S. Yajima, J. Hayashi, and K. Okamura, Nature, 266 , 521 (1977).
29. S. Yajima, Y. Hasegawa, K. Okamura, and T. Matsuzowa, Nature, 273,
525 (1978).
30. S. Yajima, J. Hayashi, M. Omori, U.S. Patent No. 4,052,430 (1978).
31. I. Taniguchi, K. Harada, and T. Maeda, Chem. Abstr., 85, 96582V
(1976).
32. R. Rice, K. Wynne, and W. Fox, U.S. Patent No. 4,097,294 (1978).
265
33. B. Walker, R. Rice, P. Becker, B. Bender, and U. Coblenz, Am.
Ceram. Soc. Bull., 62_ 916 (1983).
34. J. Wesson and T. Williaris, J. Polym. Sci.,21« 2833 (1979).
35. J. Wesson and T. Williams, J. Polym. Sci . , J^, 959 (1980).
36. J. Wesson and T. Williams, J. Polym. Sci . , J^, 65 (1981).
37. C. Schilling, J. Wesson, and T. Williams, Am. Ceram. Soc. Bull.,
_62_, 912 (1983).
38. C. Schilling, J. Wesson, and T. Williams, U.S. Patent No. 4,414,403
(1983).
39. C. Schilling, T. Williams, ACS Polymer Preprints,^, 1 (1984).
40. R. West, L. David, P. Djurovich, H. Yu , and R. Sinclair, Am. Ceram.
Soc. Bull., _62, 899 (1983).
41. R. Baney and J. Gaul, U.S. Patent No. 4,310,651 (1982).
42. R. Baney and J. Gaul, U.S. Patent No. 4,298,558 (1981).
43. R. Baney and J. Gaul, U.S. Patent No. 4,298,559 (1981).
44. R. Baney, U.S. Patent No. 4,310,481 (1982).
45. R. Baney, U.S. Patent No. 4,310,482 (1982).
46. R. Baney and J. Gaul, U.S. Patent No. 4,310,956 (1982).
47. B. Penn, F. Ledbetter, J. demons, and J. Daniels, J. Appl . Polym.
Sci., _27_, 3751 (1982).
48. D. Seyferth, G. Wiseman, and C. Prud'nomme, J. Am. Ceram. Soc.,J^
66 (1983).
49. D. Seyferth and G. Wiseman, "Silazane Precursors to Silicon Nit-
ride," in Ultrastructure Processing of Ceramics, Glasses, and Com-
posites, L. L. Hench and D. R. Ulrich, eds., John Wiley and Sons,
New York, 1984, pp. 265-271.
50. F. S. Kipping, J. Chem. Soc, 225., 2291 (1924).
51. E. G. Rochow, An Introduction to the Chemistry of the Silicons,
John Wiley and Sons, New York, 1946, p. 46.
52. E. Carberry and R. West, J. Ajti. Chem. Soc, 92, 5449 (1969).
53. S. Yajima, J. Hayashi , and M. Omori , U.S. Patent No. 4,100,233
(1978).
267
54. S. Yajima, Y. Hasegawa, J. Hayashi , and M. limura, J. Mat. Sci.,
13. 2569 (1978).
55. Y. Hasegawa, M. limura, and S. Yajima, J. Mat. Sci., _15_, 720
(1980).
56. L. David, "New Polysilane Polymers: Polysilastyrene and Poly-
phenylmethyl silylene," Ph.D. Dissertation, University of Wisconsin,
1981, p. 61.
57. B. Ankles, President of Petrarch Systems, Inc., private communica-
tion.
58. A. Miller, "Polysiloxanes ," in Radiation Chemistry of Polymers, M.
Dole and R. Salovey, eds.. Academic Press, New York, 1973, p. 179.
59. Y. Okada, "Irradiation of Polymers," in Advances in Chemistry
Series, R. F. Gould, ed.. Am. Chem. Soc, Washington, DC, 1969, p.
44.
60. J. Speier, J. Webster, and G. Barnes, J. An. Chem. Soc, 19_, 974
(1957).
61. C. Schilling, Union Carbide, private communication, 1985.
62. C. Morterra and M. J. D. Low, Chem. Commun., 23_, 203 (1968).
63. J. Roberts and M. Caserio, Basic Principles of Organic Chemistry,
2nd edition, W. A. Benzamin, Inc., Menlo Park, California, 1977, p.
92.
64. M. Kumada and K. Shiina, J. Org. Chem.,_23_, 139 (1958).
65. CRC Handbook of Chemistry and Physics, 52nd ed.. Chemical Rubber
Co., Cleveland, OH, p. F-186.
66. R. A. Sinclair, "Pyrolysis of Polysil anes ," in Ultrastructure Pro-
cessing of Ceramics, Glasses, and Composites, L. L. Hench and D. R.
Ulrich, eds., John Wiley and Sons, New York, 1984, p. 258.
67. N. A. Toropov and V. P. Barzakovskii , High-Temperature Chemistry of
Silicates and Other Oxide Systems, Consultants Bureau, New York,
1966, p. 126.
68. H. P. Kirchner, "Strengthening of Ceramics," Marcel Dekker, Inc.,
New York, 1979.
69. I. W. Donald and P. W. McMillan, J. Mat. Sci.,_n, 949 (1976).
70. A. Kelly and R. B. Robinson, Prog. Mat. Sci., 10, 1 (1963).
268
71. D. P. H. Hasselman and R. M, Fulrath, J. Am. Ceram. Soc, 49_, 68
(1966).
72. F. F. Lange, J. Am. Ceram. Soc.,_54^, 614 (1971).
73. G. R. Antis, P. Chantikul, B. R. Lawn, and D. B. Marshall, J. Am.
Ceram. Soc.,_64, 532 (1981).
74. J. K. MacKenzie, Proc. Physic. Soc. (London), B63, 2 (1950).
75. W. D. Kingery, H. K. Bowen, and D. R. Uhlmann, Introduction to
Ceramics, 2nd edition, John Wiley and Sons, New York, 1976, p. 595.
76. W. D. Kingery, H. K. Bowen, and D. R. Uhlman, Introduction to Cer-
amics, 2nd edition, John Wiley and Sons, New York, 1976, pp. 830-
844.
77. E. Fitzer and R. Ebi , "Kinetic Studies on the Oxidation of Silicon
Carbide," in Silicon Carbide-1973, R. C. Marshall, J. Faust, and C.
F. Ryan, eds.. University of South Carolina Press, Columbia, SC,
1973, p. 320.
78. F. Gac, J. W. Milewski, J. J. Petrovic, and P. D. Schalek, "Silicon
Carbide Whisker Reinforced Glass and Ceramics," in Metal Matrix,
Carbon, and Ceramic Matrix Composites 1985, NASA Conference Publi-
cation 2406, Hampton, VA, J. D. Buckley, ed., p. 53.
79. S. C. Samanta and S. Musikant, "SiC Whiskers-Reinforced Ceramic
Matrix Composites," in Ceramic Eng. & Sci., Proc, Am. Ceramic
Soc, Columbus, OH, July-Aug. 1985, p. 663.
80. F. P. Meyer, G. D. Quinn, and J. C. Walk, "Reinforcing Fused Silica
with High Purity Fibers" in Ceramic Eng. St Sci., Am. Ceramic Soc,
Columbus, OH, July-Aug. 1985, p. 646.
81. R. A. Cutler, A. V. Virkar, and J. Holt, "Synthesis and Densifica-
tion of Oxide-Carbide Composites," in Ceramic Eng. & Sci., Am.
Ceramic Soc, Columbus, OH, July-Aug. 1985, p. 715.
82. F. F. Lange, B. I. Davis, and E. Wright, J. Am. Ceram. Soc, 69 , 66
(1986).
83. J. Zarzycki , M. Parassas, and J. Phalippou, J. Mat. Sci.jJJ, 3371
(1982).
84. P. Yu, H. Liu, and Y. Wang, J. Non-Cryst. Solids, _5^, 511 (1982).
85. A. V. Kiselev and V. I. Lygin, Infrared Spectra of Surface Com-
pounds, Halsted Press, New York, 1975, p. 117.
269
86. K. Hashimoto, K. Asami , and K. Teramoto, Corrosion Sci., _19., 3
(1978).
87. S. Wallace and L. L. Hench, "The Processing and Characterization
of DCCA Modified Gel -Derived Silica" in Better Ceramics Through
Chemistry, C. J. Brinker, D. E. Clark, and D. R. Ulrich, eds..
Material Research Society Symposium Proceeding, Vol. 32, North-
Holland Pub. Co., New York, NY, 1984, p. 47.
88. R. A. J. Sambell, A. Briggs, D. C. Phillips, and D. H. Bowen, J.
Mater. Sci., 1_, 676 (1972).
89. D. C. Phillips, J. Mater. Sci.,_7_, 1175 (1972).
90. D. C. Phillips, J. Mater. Sci.,_9_, 1847 (1974).
91. K. M. Prewo and J. J. Brennan, J. Mater. Sci. \b_, 463 (1980).
92. K. M, Prewo and J. J.. Brennan, J. Mater. Sci. 21, 1201 (1982).
93. K. M. Prewo and J. J. Brennan, J. Mater. Sci. r7_, 2371 (1982).
94. K. M. Prewo, J. Mater. Sci._lZ, 3549 (1982).
95. W. W. Pultz, J. Am. Ceramic Soc.,^, 419 (1966).
96. P. J. Jorgensen, M. E. Wadsworth, and I. R. Cutler, J. Am. Ceramic
Soc, _44, 258 (1961).
97. G. E. Ervin, Jr., J. Am. Ceramic Soc.,_4]_, 347 (1958).
98. J. A. Costello and R. E. Tressler, J. Am. Ceramic Soc, _64_, 327
(1981).
99. S. C. Singhal, J. Materials Sci., 21, 1246 (1976).
100. E. A. Gulbransen and S. A. Jansson, Oxid. Met.,_4. 181 (1981).
101. E. A. Gulbransen, K. R. Andrew, and F. A. Brassart, J.
Electrochem. Soc. 113, 1311 (1966).
102. J. Bialoskorski and K. Konsztowicz, Szklo Ceram.,_28, 207 (1977).
103. F. E. Wang and K. A. Sutula, J. Appl . Phys., 49_, 944 (1978).
104. J. J. Lannutti and D. E. Clark, "Sol-Gel Derived Coatings on SiC
and Silicate Fibers" in Ceramic Proceedings, 8th Annual Conference
on Composites and Advanced Ceramic Material, Am. Ceram. Soc,
Columbus, Ohio, 1984, p. 574.
270
105. J. J. Lannutti and D. E. Clark, "Sol-Gel Derived Ceramic-Ceramic
Composites" in Better Ceramics Through Chemistry. C. J. Brinker,
D. E. Clark, and D. R. Ulrich, eds.. Materials Research Society
Symposia Proceedings, Vol. 32, North-Holland, New York, 1984, p.
369.
106. J. J. Lannutti and D. E. Clark, "Long Fiber Reinforced Sol-Gel
Derived Al 003 in Better Ceramics Through Chemistry. C. J. Brinker,
D. E. Clark, and D. R. Ulrich, eds.. Materials Research Society
Symposia Proceedings, Vol. 32, North-Holland, New York, 1984, p.
375.
107. R. W. Rice, J. R. Span, D. Lewis, and W. Coblenz, "Effect of
Ceramic Fiber Coatings on the Room-Temperature Mechanical Behavior
of Ceramic-Fiber Composites" in Ceramic Proceedings, 8th Annual
Conference on Composites and Advanced Ceramic Materials, Am.
Ceram. Soc, Columbus, Ohio, 1984, p. 614.
108. Annual Book of ASTM Standards, "Flexural Testing of Glass," sec-
tion 15, Am. Soc. of Testing Mater., Philadelphia, PA, 1983, p.
25.
109. G. K. Bansal and W. H. Duckworth, "Fracture Surface Energy
Measurements by the Notch-Beam Technique," in Fracture Mechanics
Applied to Brittle Materials, S. W. Freiman, ed. , ASTM STP 678,
Am. Soc. for Testing and Materials, Philadelphia, PA, 1978, p. 38.
110. H. H. Willard, L. L. Merritt, J. A. Dean, and F. A. Settle, In-
strumental Methods of Analysis, 6th ed., D. Van Nostrand Co., New
York, 19R1, p. 209.
111. A. K. Ramdas, Indian Academy of Sci., proceedings, 37AA , 571
(1953).
112. M. Haccuria, Bull. Soc. Chim., Belgium, _62_, 428 (1953).
113. I. Simm and H. 0. McMahon, J. of Chem. Phys.,_2]_23 (1953).
114. R. A. J. Sambel 1 , D. H. Bowen, and D. C. Phillips, J. Mater. Sci.,
2, 663 (1972).
115. J. L. Henshall , D. J. Rowel iffe, and J. W. Eddington, J. Am.
Ceram. Soc.,_60_, 373 (1977).
116. J. L. Henshall and C. A. Brookes, J. Mater. Sci. Letters, _4, 783
(1985).
117. G. C. Wei and P. F. Becher, J. Am. Cer. Soc. , _67_, 571 (1984).
118. S. R. Nutt and F. E. Wawner, J. Mater. Sci., 20, 1953 (1985).
271
119. B. A. Bender, D. Lewis, W. S. Coblenz, and R. W. Rice, "Electron
Microscopy of Ceramic Fiber-Ceramic Matrix Composites - Comparison
with Processing and Behavior," in NASA Conference Pub. 2357,
Hampton, VA, J. D. Buckley, ed., 1984, p. 171.
120. T. J. Clark, R. M. Arons, J. B. Stamatoff, and J. Rabe, "Thermal
Degradation of Nicalon" SiC Fibers," in Ceramic Eng. & Sci. Proc,
Vol. 6, Am. Ceramic Soc, Columbus, Ohio, 1985, p. 576.
121. R. E. Wilson and D. M Breit, "Near Net Shape SiC Whisker Reinforc-
ed SiC Oxidation Resistant Structural Composites Made by the
TEOSiC Process," in Metal Matrix, Carbon, and Ceramic Matrix Com-
posites 1985, J. D. Buckley, ed., p. 229.
122. R. Rice, "Ceramic Composites," in Ceramic Eng. & Sci. Proc, Vol.
2, Am. Ceramic. Soc, Columbus, Ohio, 1981, p. 493.
123. G. C. Wei and P. F. Becher, Am. Ceram. Bull., 64, 298 (1985).
124. R. Pampuch, W. S. Ptak, S. Jonas, and S. Stoch, Proc 4th CIMTES,
St. Vincent, Amy, 1974, P. Vincenzini, ed., Elsevier Pub. Co.,
Amsterdam, 1980.
125. V. A. Lavenko, S. Jones, and R. Panpuch, Ceramic International, 7,
75 (1981). ~
126. S. C. Singhal and F. F. Lange, J. Am. Ceram. Soc. , _58_, 433 (1975).
127. A. Suzuki, H. Ashido, N. Fruii, K. Mameno, and H. Matsunami ,
Japan, J. Appl . Phys.,_2]_, 579 (1982).
128. S. Y. Lee, "Low Temperature Oxidation of SiC," Third Annual Report
to Air Force Office of Scientific Research, L. L. Hench and D. E.
Clark, eds., Gainesville, Florida, March 1982-Feb. 1983, p. 158.
129. G. P. LaTorre, R. A. Stokell, R. H. Krabill, and D. E. Clark, Air
Force Report from University of Florida, April 1983-March 1985.
To be presented at the Engineering Ceramic Conf., Cocoa Beach, FL,
Jan. 1986.
BIOGRAPHICAL SKETCH
Burtrand Insung Lee attended and yraduated from Montgomery Blair
High School in Silver Spring, Maryland and Southern College in
Collegedale, Tennessee, in 1971 and 1975, respectively. Upon graduation
from Southern College as Chemistry major, he was employed oy Biospherics
Inc., in Rockville, Maryland, as a chemist. At Biospherics he was in-
directly engaged in the Viking Project search of life on Mars.
After the decision to pursue full-time graduate work, he enrolled
and graduated from Western Michigan University majoring in analytical
chemistry in 1979. After some experiences in teaching and research in
chemistry at State University of Mew York, he entered the University of
Florida as a doctoral student.
272
I certify that I have read this study and that in my opinion it
conforms to acceptable standards of scholarly presentation and is fully
adequate in scope and quality, as a dissertation for the degree of
Doctor of Philosophy.
Larry L. Hench, Chairman
Professor of Materials
and Engineering
Science
I certify that I have read this study and that in my opinion it
conforms to acceptable standards of scholarly presentation and is fully
adequate in scope and quality, as a dissertation for the degree of
Doctor of Philosophy.
Christopher D. Batich
Associate Professor of Materials
Science and Engineering
I certify that I have read this study and that in my opinion it
conforms to acceptable standards of scholarly presentation and is fully
adequate in scope and quality, as a dissertation for the degree of
Doctor of Philosophy.
Mr UhCL,
Michael D. Sacks
Associate Professor of Materials
Science and Engineering
I certify that I have read this study and that in my opinion it
conforms to acceptable standards of scholarly presentation and is fully
adequate in scope and quality, as a dissertation for the degree of
Doctor of Philosophy.
Lawrence E. Malvern
Professor of Engineering Science
I certify that I have read this study and that in my opinion it
conforms to acceptable standards of scholarly presentation and is fully
adequate in scope and quality, as a dissertation for the degree of
Doctor of Philosophy.
David E. Clark
Associate Professor of Materials
Science and Engineering
This dissertation was submitted to the Graduate Faculty of the College
of Engineering and to the Graduate School, and was accepted as partial
fulfillment of the requirements for the degree of Doctor of Philosophy.
May 1986
Dean, Graduate School
TVNews
OpenLibrary